add run results and update software versions

2026-08-15 19:38:27 +00:00 · 2024-10-11 10:32:29 -04:00
342 changed files with 13906 additions and 85936 deletions
@@ -1,7 +1,6 @@
 name: Benchmarks
 env:
  # TODO: this rescheduling makes gpt2, mixtral and llama unjitted slower
-  # TODO: very slow for llama 70B and resnet training 6 GPU
  RUN_PROCESS_REPLAY: "1"
  ASSERT_PROCESS_REPLAY: "0"
  PYTHONPATH: .
@@ -25,7 +24,6 @@ jobs:
  testmacbenchmark:
    name: Mac Benchmark
    runs-on: [self-hosted, macOS]
-    timeout-minutes: 20
    defaults:
      run:
        shell: bash -o pipefail {0}
@@ -47,59 +45,55 @@ jobs:
        echo "CACHEDB=/tmp/staging.db" >> $GITHUB_ENV
        rm -f /tmp/staging.db /tmp/staging.db-shm /tmp/staging.db-wal
    - name: reset process replay
-      run: python3.11 test/external/process_replay/reset.py
+      run: test/external/process_replay/reset.py
    - name: Run Stable Diffusion
-      run: JIT=1 python3.11 examples/stable_diffusion.py --fp16 --seed 0 --noshow --timing | tee sd.txt
-    - name: Run Stable Diffusion without fp16
-      run: JIT=1 python3.11 examples/stable_diffusion.py --seed 0 --noshow --timing | tee sd_no_fp16.txt
-    - name: Run Stable Diffusion v2
-      run: JIT=1 python3.11 examples/sdv2.py --fp16 --seed 0 --noshow --timing | tee sdv2.txt
+      run: JIT=2 python3 examples/stable_diffusion.py --seed 0 --noshow --timing | tee sd.txt
+    - name: Run Stable Diffusion with fp16
+      run: JIT=2 python3 examples/stable_diffusion.py --fp16 --seed 0 --noshow --timing | tee sd_fp16.txt
    - name: Run SDXL
-      run: JIT=1 python3.11 examples/sdxl.py --seed 0 --noshow --timing | tee sdxl.txt
+      run: JIT=2 python3 examples/sdxl.py --seed 0 --noshow --timing | tee sdxl.txt
    - name: Run model inference benchmark
-      run: METAL=1 python3.11 test/external/external_model_benchmark.py
+      run: METAL=1 python3 test/external/external_model_benchmark.py
    - name: Test speed vs torch
-      run: BIG=2 MPS=1 python3.11 test/test_speed_v_torch.py | tee torch_speed.txt
+      run: BIG=2 MPS=1 python3 test/test_speed_v_torch.py | tee torch_speed.txt
    - name: Test tensor cores
-      run: METAL=1 python3.11 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded
-    - name: Run Tensor Core GEMM (float)
-      run: DEBUG=2 python3.11 extra/gemm/simple_matmul.py | tee matmul.txt
-    - name: Run Tensor Core GEMM (half)
-      run: DEBUG=2 HALF=1 python3.11 extra/gemm/simple_matmul.py | tee matmul_half.txt
-    - name: Run Tensor Core GEMM (bfloat16)
-      run: DEBUG=2 BFLOAT16=1 python3.11 extra/gemm/simple_matmul.py | tee matmul_bfloat16.txt
+      run: METAL=1 python3 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded
+    - name: Run Tensor Core GEMM
+      run: |
+        DEBUG=2 python3 extra/gemm/simple_matmul.py | tee matmul.txt
+        DEBUG=2 HALF=1 python3 extra/gemm/simple_matmul.py | tee matmul_half.txt
    - name: Fuzz Padded Tensor Core GEMM
-      run: METAL=1 M_START=6 M_STOP=10 M_STEP=1 N_START=6 N_STOP=10 N_STEP=1 K_START=6 K_STOP=24 K_STEP=1 TC_OPT=2 DEBUG=2 python3.11 ./extra/gemm/fuzz_matmul.py
+      run: METAL=1 M_START=6 M_STOP=10 M_STEP=1 N_START=6 N_STOP=10 N_STEP=1 K_START=6 K_STOP=24 K_STEP=1 TC_OPT=2 DEBUG=2 python3 ./extra/gemm/fuzz_matmul.py
    - name: Run LLaMA
      run: |
-        JIT=0 python3.11 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_unjitted.txt
-        JIT=1 python3.11 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_jitted.txt
+        JIT=0 python3 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_unjitted.txt
+        JIT=1 python3 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_jitted.txt
    - name: Run LLaMA with BEAM
-      run: JITBEAM=2 IGNORE_BEAM_CACHE=1 python3.11 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_beam.txt
+      run: JITBEAM=2 IGNORE_BEAM_CACHE=1 python3 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_beam.txt
    - name: Run quantized LLaMA
      run: |
-        python3.11 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing --quantize int8 | tee llama_int8.txt
-        python3.11 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing --quantize nf4 | tee llama_nf4.txt
+        python3 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing --quantize int8 | tee llama_int8.txt
+        python3 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing --quantize nf4 | tee llama_nf4.txt
    - name: Run LLaMA 7B on 4 (virtual) GPUs
-      run: python3.11 examples/llama.py --gen 1 --size 7B --shard 4 --prompt "Hello." --count 10 --temperature 0  --timing | tee llama_four_gpu.txt
+      run: python3 examples/llama.py --gen 1 --size 7B --shard 4 --prompt "Hello." --count 10 --temperature 0  --timing | tee llama_four_gpu.txt
    - name: Run GPT2
      run: |
-        JIT=0 python3.11 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_unjitted.txt
-        JIT=1 python3.11 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_jitted.txt
+        JIT=0 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_unjitted.txt
+        JIT=1 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_jitted.txt
    - name: Run GPT2 w HALF
-      run: HALF=1 python3.11 examples/gpt2.py --count 10 --temperature 0 --timing | tee gpt2_half.txt
+      run: HALF=1 python3 examples/gpt2.py --count 10 --temperature 0 --timing | tee gpt2_half.txt
    - name: Run GPT2 w HALF/BEAM
-      run: HALF=1 JITBEAM=2 IGNORE_BEAM_CACHE=1 CAST_BEFORE_VIEW=0 python3.11 examples/gpt2.py --count 10 --temperature 0 --timing | tee gpt2_half_beam.txt
+      run: HALF=1 JITBEAM=2 IGNORE_BEAM_CACHE=1 CAST_BEFORE_VIEW=0 python3 examples/gpt2.py --count 10 --temperature 0 --timing | tee gpt2_half_beam.txt
    - name: Train MNIST
-      run: time PYTHONPATH=. TARGET_EVAL_ACC_PCT=96.0 python3.11 examples/beautiful_mnist.py | tee beautiful_mnist.txt
+      run: time PYTHONPATH=. TARGET_EVAL_ACC_PCT=97.3 python3 examples/beautiful_mnist.py | tee beautiful_mnist.txt
    - name: Run 10 CIFAR training steps
-      run: JIT=1 STEPS=10 python3.11 examples/hlb_cifar10.py | tee train_cifar.txt
+      run: JIT=2 STEPS=10 python3 examples/hlb_cifar10.py | tee train_cifar.txt
    - name: Run 10 CIFAR training steps w HALF
-      run: JIT=2 STEPS=10 DEFAULT_FLOAT=HALF python3.11 examples/hlb_cifar10.py | tee train_cifar_half.txt
+      run: JIT=2 STEPS=10 DEFAULT_FLOAT=HALF python3 examples/hlb_cifar10.py | tee train_cifar_half.txt
    #- name: Run 10 CIFAR training steps w BF16
-    #  run: STEPS=10 DEFAULT_FLOAT=BFLOAT16 python3.11 examples/hlb_cifar10.py | tee train_cifar_bf16.txt
+    #  run: STEPS=10 DEFAULT_FLOAT=BFLOAT16 python3 examples/hlb_cifar10.py | tee train_cifar_bf16.txt
    - name: Run 10 CIFAR training steps w winograd
-      run: JIT=1 WINO=1 STEPS=10 python3.11 examples/hlb_cifar10.py | tee train_cifar_wino.txt
+      run: JIT=2 WINO=1 STEPS=10 python3 examples/hlb_cifar10.py | tee train_cifar_wino.txt
    - uses: actions/upload-artifact@v4
      with:
        name: Speed (Mac)
@@ -118,10 +112,8 @@ jobs:
          gpt2_half_beam.txt
          matmul.txt
          matmul_half.txt
-          matmul_bfloat16.txt
          sd.txt
-          sd_no_fp16.txt
-          sdv2.txt
+          sd_fp16.txt
          sdxl.txt
          beautiful_mnist.txt
          train_cifar.txt
@@ -129,12 +121,11 @@ jobs:
          train_cifar_bf16.txt
          train_cifar_wino.txt
    - name: Run process replay tests
-      run: cp test/external/process_replay/process_replay.py ./process_replay.py && git fetch origin master && git -c advice.detachedHead=false checkout origin/master && PYTHONPATH=. python3.11 process_replay.py
+      run: cp test/external/process_replay/process_replay.py ./process_replay.py && git fetch origin master && git -c advice.detachedHead=false checkout origin/master && PYTHONPATH=. python3 process_replay.py

  testnvidiabenchmark:
    name: tinybox green Benchmark
    runs-on: [self-hosted, Linux, tinyboxgreen]
-    timeout-minutes: 20
    defaults:
      run:
        shell: bash -o pipefail {0}
@@ -161,11 +152,9 @@ jobs:
    - name: reset process replay
      run: test/external/process_replay/reset.py
    - name: Run model inference benchmark
-      run: NV=1 RUN_PROCESS_REPLAY=0 NOCLANG=1 python3 test/external/external_model_benchmark.py
+      run: NV=1 NOCLANG=1 python3 test/external/external_model_benchmark.py
    - name: Test speed vs torch
-      run: NV=1 RUN_PROCESS_REPLAY=0 HALF=1 BIG=2 TORCHCUDA=1 python3 test/test_speed_v_torch.py | tee torch_speed.txt
-    - name: Test speed vs theoretical
-      run: NV=1 IGNORE_BEAM_CACHE=1 BEAM_DEBUG=1 DEBUG=1 python -m pytest -rA test/external/speed_v_theoretical.py --durations=20
+      run: NV=1 BIG=2 TORCHCUDA=1 python3 test/test_speed_v_torch.py | tee torch_speed.txt
    - name: Test tensor cores
      run: |
        NV=1 python3 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded
@@ -178,34 +167,32 @@ jobs:
      run: NV=1 PTX=1 HALF=1 DEBUG=2 python3 extra/gemm/simple_matmul.py | tee matmul_ptx.txt
    - name: Run Tensor Core GEMM (NV)
      run: NV=1 HALF=1 DEBUG=2 python3 extra/gemm/simple_matmul.py | tee matmul_nv.txt
-    - name: Test NV=1
-      run: DEBUG=2 NV=1 python -m pytest -rA test/test_tiny.py
-    - name: Test CUDA=1
-      run: DEBUG=2 CUDA=1 python -m pytest -rA test/test_tiny.py
+    - name: Run Tensor Core GEMM (NV) with BEAM
+      run: BEAM=4 NV=1 HALF=1 IGNORE_BEAM_CACHE=1 DEBUG=2 python3 extra/gemm/simple_matmul.py
    - name: Run Stable Diffusion
-      run: NV=1 python3 examples/stable_diffusion.py --fp16 --seed 0 --noshow --timing | tee sd.txt
+      run: NV=1 python3 examples/stable_diffusion.py --seed 0 --noshow --timing | tee sd.txt
    - name: Run SDXL
-      run: NV=1 RUN_PROCESS_REPLAY=0 python3 examples/sdxl.py --seed 0 --noshow --timing | tee sdxl.txt
+      run: NV=1 python3 examples/sdxl.py --seed 0 --noshow --timing | tee sdxl.txt
    - name: Run LLaMA
      run: |
        NV=1 JIT=0 python3 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_unjitted.txt
        NV=1 JIT=1 python3 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_jitted.txt
    - name: Run LLaMA with BEAM
      run: NV=1 JITBEAM=2 IGNORE_BEAM_CACHE=1 python3 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_beam.txt
-    # - name: Run LLaMA 7B on 4 GPUs
-    #   run: NV=1 RUN_PROCESS_REPLAY=0 python3 examples/llama.py --gen 1 --size 7B --shard 4 --prompt "Hello." --count 10 --temperature 0  --timing | tee llama_four_gpu.txt
-    # - name: Run LLaMA 7B on 6 GPUs
-    #   run: NV=1 RUN_PROCESS_REPLAY=0 python3 examples/llama.py --gen 1 --size 7B --shard 6 --prompt "Hello." --count 10 --temperature 0  --timing | tee llama_six_gpu.txt
+    - name: Run LLaMA 7B on 4 GPUs
+      run: NV=1 python3 examples/llama.py --gen 1 --size 7B --shard 4 --prompt "Hello." --count 10 --temperature 0  --timing | tee llama_four_gpu.txt
+    - name: Run LLaMA 7B on 6 GPUs
+      run: NV=1 python3 examples/llama.py --gen 1 --size 7B --shard 6 --prompt "Hello." --count 10 --temperature 0  --timing | tee llama_six_gpu.txt
    - name: Run LLaMA-3 8B BEAM
-      run: NV=1 JITBEAM=2 IGNORE_BEAM_CACHE=1 python3 examples/llama3.py --size 8B --model weights/LLaMA-3/8B-SF-DPO/ --benchmark --temperature 0 | tee llama3_beam.txt
+      run: NV=1 JITBEAM=2 IGNORE_BEAM_CACHE=1 python3 examples/llama3.py --model weights/LLaMA-3/8B-SF-DPO/ --benchmark --temperature 0 | tee llama3_beam.txt
    - name: Run LLaMA-3 8B on 4 GPUs
-      run: NV=1 RUN_PROCESS_REPLAY=0 python3 examples/llama3.py --size 8B --shard 4 --model weights/LLaMA-3/8B-SF-DPO/ --benchmark --temperature 0 | tee llama3_four_gpu.txt
+      run: NV=1 python3 examples/llama3.py --shard 4 --model weights/LLaMA-3/8B-SF-DPO/ --benchmark --temperature 0 | tee llama3_four_gpu.txt
    - name: Run LLaMA-3 8B on 6 GPUs
-      run: NV=1 RUN_PROCESS_REPLAY=0 python3 examples/llama3.py --size 8B --shard 6 --model weights/LLaMA-3/8B-SF-DPO/ --benchmark --temperature 0 | tee llama3_six_gpu.txt
+      run: NV=1 python3 examples/llama3.py --shard 6 --model weights/LLaMA-3/8B-SF-DPO/ --benchmark --temperature 0 | tee llama3_six_gpu.txt
    - name: Run LLaMA-2 70B
-      run: NV=1 RUN_PROCESS_REPLAY=0 MAX_CONTEXT=256 python3 examples/llama.py --gen 2 --size 70B --shard 6 --prompt "Hello." --count 10 --temperature 0  --timing | tee llama_2_70B.txt
+      run: NV=1 MAX_CONTEXT=256 python3 examples/llama.py --gen 2 --size 70B --shard 6 --prompt "Hello." --count 10 --temperature 0  --timing | tee llama_2_70B.txt
    - name: Run Mixtral 8x7B
-      run: time NV=1 RUN_PROCESS_REPLAY=0 python3 examples/mixtral.py --temperature 0 --count 10 --timing | tee mixtral.txt
+      run: time NV=1 python3 examples/mixtral.py --temperature 0 --count 10 --timing | tee mixtral.txt
    - name: Run GPT2
      run: |
        NV=1 JIT=0 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_unjitted.txt
@@ -229,6 +216,8 @@ jobs:
          llama_unjitted.txt
          llama_jitted.txt
          llama_beam.txt
+          llama_four_gpu.txt
+          llama_six_gpu.txt
          llama3_beam.txt
          llama3_four_gpu.txt
          llama3_six_gpu.txt
@@ -244,7 +233,6 @@ jobs:
  testmorenvidiabenchmark:
    name: tinybox green Training Benchmark
    runs-on: [self-hosted, Linux, tinyboxgreen]
-    timeout-minutes: 20
    defaults:
      run:
        shell: bash -o pipefail {0}
@@ -274,7 +262,7 @@ jobs:
    - name: Fuzz Padded Tensor Core GEMM (PTX)
      run: NV=1 PTX=1 M_START=12 M_STOP=20 M_STEP=1 N_START=6 N_STOP=10 N_STEP=1 K_START=28 K_STOP=36 K_STEP=1 HALF=1 TC_OPT=2 python3 ./extra/gemm/fuzz_matmul.py
    - name: Train MNIST
-      run: time PYTHONPATH=. NV=1 TARGET_EVAL_ACC_PCT=96.0 python3 examples/beautiful_mnist.py | tee beautiful_mnist.txt
+      run: time PYTHONPATH=. NV=1 TARGET_EVAL_ACC_PCT=97.3 python3 examples/beautiful_mnist.py | tee beautiful_mnist.txt
    - name: Run 10 CIFAR training steps
      run: NV=1 STEPS=10 python3 examples/hlb_cifar10.py | tee train_cifar.txt
    - name: Run 10 CIFAR training steps w HALF
@@ -282,17 +270,17 @@ jobs:
    - name: Run 10 CIFAR training steps w BF16
      run: NV=1 STEPS=10 DEFAULT_FLOAT=BFLOAT16 python3 examples/hlb_cifar10.py | tee train_cifar_bf16.txt
    - name: Run 10 CIFAR training steps w winograd
-      run: NV=1 RUN_PROCESS_REPLAY=0 WINO=1 STEPS=10 DEFAULT_FLOAT=HALF python3 examples/hlb_cifar10.py | tee train_cifar_wino.txt
+      run: NV=1 WINO=1 STEPS=10 python3 examples/hlb_cifar10.py | tee train_cifar_wino.txt
    - name: Run full CIFAR training w 1 GPU
      run: time NV=1 DEFAULT_FLOAT=HALF LATEWINO=1 STEPS=1000 TARGET_EVAL_ACC_PCT=93.2 python3 examples/hlb_cifar10.py | tee train_cifar_one_gpu.txt
    - name: Run full CIFAR training steps w 6 GPUS
-      run: time RUN_PROCESS_REPLAY=0 NV=1 DEFAULT_FLOAT=HALF STEPS=350 BS=1536 GPUS=6 TARGET_EVAL_ACC_PCT=93.2 python3 examples/hlb_cifar10.py | tee train_cifar_six_gpu.txt
+      run: time NV=1 DEFAULT_FLOAT=HALF STEPS=350 BS=1536 GPUS=6 TARGET_EVAL_ACC_PCT=93.2 python3 examples/hlb_cifar10.py | tee train_cifar_six_gpu.txt
    - name: Run MLPerf resnet eval on training data
      run: time NV=1 MODEL=resnet python3 examples/mlperf/model_eval.py
    - name: Run 10 MLPerf ResNet50 training steps (1 gpu)
      run: NV=1 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=256 GPUS=1 MODEL=resnet python3 examples/mlperf/model_train.py | tee train_resnet_one_gpu.txt
    - name: Run 10 MLPerf ResNet50 training steps (6 gpu)
-      run: NV=1 RUN_PROCESS_REPLAY=0 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=1536 GPUS=6 MODEL=resnet python3 examples/mlperf/model_train.py | tee train_resnet.txt
+      run: NV=1 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=1536 GPUS=6 MODEL=resnet python3 examples/mlperf/model_train.py | tee train_resnet.txt
    - uses: actions/upload-artifact@v4
      with:
        name: Speed (NVIDIA Training)
@@ -312,7 +300,6 @@ jobs:
  testamdbenchmark:
    name: tinybox red Benchmark
    runs-on: [self-hosted, Linux, tinybox]
-    timeout-minutes: 20
    defaults:
      run:
        shell: bash -o pipefail {0}
@@ -338,10 +325,6 @@ jobs:
        rm -f /tmp/staging.db /tmp/staging.db-shm /tmp/staging.db-wal
    - name: reset process replay
      run: test/external/process_replay/reset.py
-    - name: setup perflevel
-      run: |
-        examples/mlperf/training_submission_v4.1/tinycorp/benchmarks/bert/implementations/tinybox_red/setup.sh
-        rocm-smi
    - name: Show off tinybox
      run: /opt/rocm/bin/rocm-bandwidth-test
    # TODO: unstable on AMD
@@ -352,21 +335,16 @@ jobs:
    #  run: |
    #    python3 -c "import torch; print(torch.__version__)"
    #    LD_PRELOAD="/opt/rocm/lib/libhsa-runtime64.so" HSA=1 BIG=2 TORCHCUDA=1 python3 test/test_speed_v_torch.py | tee torch_speed.txt
-    - name: Test speed vs theoretical
-      run: AMD=1 IGNORE_BEAM_CACHE=1 BEAM_DEBUG=1 DEBUG=1 python -m pytest -rA test/external/speed_v_theoretical.py --durations=20
    - name: Test tensor cores
-      run: AMD=1 python3 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded
+      run: |
+        AMD=1 python3 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded
    - name: Run Tensor Core GEMM (AMD)
      run: AMD=1 HALF=1 DEBUG=2 python3 extra/gemm/simple_matmul.py | tee matmul_amd.txt
-    - name: Test AMD=1
-      run: DEBUG=2 AMD=1 python -m pytest -rA test/test_tiny.py
-    - name: Test HIP=1
-      run: DEBUG=2 HIP=1 python -m pytest -rA test/test_tiny.py
    # TODO: AMD compiler bug causes this to fail
    #- name: Fuzz Padded Tensor Core GEMM
    #  run: HSA=1 M_START=12 M_STOP=20 M_STEP=1 N_START=12 N_STOP=20 N_STEP=1 K_START=28 K_STOP=36 K_STEP=1 HALF=1 TC_OPT=2 DEBUG=2 python3 ./extra/gemm/fuzz_matmul.py
    - name: Run Stable Diffusion
-      run: AMD=1 python3 examples/stable_diffusion.py --fp16 --seed 0 --noshow --timing | tee sd.txt
+      run: AMD=1 python3 examples/stable_diffusion.py --seed 0 --noshow --timing | tee sd.txt
    - name: Run SDXL
      run: AMD=1 python3 examples/sdxl.py --seed 0 --noshow --timing | tee sdxl.txt
    - name: Run LLaMA 7B
@@ -375,18 +353,18 @@ jobs:
        AMD=1 JIT=1 python3 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_jitted.txt
    - name: Run LLaMA 7B with BEAM
      run: AMD=1 JITBEAM=2 IGNORE_BEAM_CACHE=1 python3 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_beam.txt
-    # - name: Run LLaMA 7B on 4 GPUs
-    #   run: AMD=1 RUN_PROCESS_REPLAY=0 python3 examples/llama.py --gen 1 --size 7B --shard 4 --prompt "Hello." --count 10 --temperature 0  --timing | tee llama_four_gpu.txt
-    # - name: Run LLaMA 7B on 6 GPUs
-    #   run: AMD=1 RUN_PROCESS_REPLAY=0 python3 examples/llama.py --gen 1 --size 7B --shard 6 --prompt "Hello." --count 10 --temperature 0  --timing | tee llama_six_gpu.txt
+    - name: Run LLaMA 7B on 4 GPUs
+      run: AMD=1 python3 examples/llama.py --gen 1 --size 7B --shard 4 --prompt "Hello." --count 10 --temperature 0  --timing | tee llama_four_gpu.txt
+    - name: Run LLaMA 7B on 6 GPUs
+      run: AMD=1 python3 examples/llama.py --gen 1 --size 7B --shard 6 --prompt "Hello." --count 10 --temperature 0  --timing | tee llama_six_gpu.txt
    - name: Run LLaMA-3 8B BEAM
-      run: AMD=1 JITBEAM=2 IGNORE_BEAM_CACHE=1 python3 examples/llama3.py --size 8B --model weights/LLaMA-3/8B-SF-DPO/ --benchmark --temperature 0 | tee llama3_beam.txt
+      run: AMD=1 JITBEAM=2 IGNORE_BEAM_CACHE=1 python3 examples/llama3.py --model weights/LLaMA-3/8B-SF-DPO/ --benchmark --temperature 0 | tee llama3_beam.txt
    - name: Run LLaMA-3 8B on 4 GPUs
-      run: AMD=1 RUN_PROCESS_REPLAY=0 python3 examples/llama3.py --size 8B --shard 4 --model weights/LLaMA-3/8B-SF-DPO/ --benchmark --temperature 0 | tee llama3_four_gpu.txt
+      run: AMD=1 python3 examples/llama3.py --shard 4 --model weights/LLaMA-3/8B-SF-DPO/ --benchmark --temperature 0 | tee llama3_four_gpu.txt
    - name: Run LLaMA-3 8B on 6 GPUs
-      run: AMD=1 RUN_PROCESS_REPLAY=0 python3 examples/llama3.py --size 8B --shard 6 --model weights/LLaMA-3/8B-SF-DPO/ --benchmark --temperature 0 | tee llama3_six_gpu.txt
+      run: AMD=1 python3 examples/llama3.py --shard 6 --model weights/LLaMA-3/8B-SF-DPO/ --benchmark --temperature 0 | tee llama3_six_gpu.txt
    - name: Run LLaMA-2 70B
-      run: AMD=1 RUN_PROCESS_REPLAY=0 python3 examples/llama.py --gen 2 --size 70B --shard 6 --prompt "Hello." --count 10 --temperature 0  --timing | tee llama_2_70B.txt
+      run: AMD=1 python3 examples/llama.py --gen 2 --size 70B --shard 6 --prompt "Hello." --count 10 --temperature 0  --timing | tee llama_2_70B.txt
    - name: Run Mixtral 8x7B
      run: time AMD=1 python3 examples/mixtral.py --temperature 0 --count 10 --timing | tee mixtral.txt
    - name: Run GPT2
@@ -406,6 +384,8 @@ jobs:
          llama_unjitted.txt
          llama_jitted.txt
          llama_beam.txt
+          llama_four_gpu.txt
+          llama_six_gpu.txt
          llama3_beam.txt
          llama3_four_gpu.txt
          llama3_six_gpu.txt
@@ -425,7 +405,6 @@ jobs:
  testmoreamdbenchmark:
    name: tinybox red Training Benchmark
    runs-on: [self-hosted, Linux, tinybox]
-    timeout-minutes: 20
    defaults:
      run:
        shell: bash -o pipefail {0}
@@ -450,12 +429,8 @@ jobs:
        rm -f /tmp/staging.db /tmp/staging.db-shm /tmp/staging.db-wal
    - name: reset process replay
      run: test/external/process_replay/reset.py
-    - name: setup perflevel
-      run: |
-        examples/mlperf/training_submission_v4.1/tinycorp/benchmarks/bert/implementations/tinybox_red/setup.sh
-        rocm-smi
    - name: Train MNIST
-      run: time PYTHONPATH=. AMD=1 TARGET_EVAL_ACC_PCT=96.0 python3 examples/beautiful_mnist.py | tee beautiful_mnist.txt
+      run: time PYTHONPATH=. AMD=1 TARGET_EVAL_ACC_PCT=97.3 python3 examples/beautiful_mnist.py | tee beautiful_mnist.txt
    - name: Run 10 CIFAR training steps
      run: AMD=1 STEPS=10 python3 examples/hlb_cifar10.py | tee train_cifar.txt
    - name: Run 10 CIFAR training steps w HALF
@@ -463,7 +438,7 @@ jobs:
    - name: Run 10 CIFAR training steps w BF16
      run: AMD=1 STEPS=10 DEFAULT_FLOAT=BFLOAT16 python3 examples/hlb_cifar10.py | tee train_cifar_bf16.txt
    - name: Run 10 CIFAR training steps w winograd
-      run: AMD=1 WINO=1 STEPS=10 DEFAULT_FLOAT=HALF python3 examples/hlb_cifar10.py | tee train_cifar_wino.txt
+      run: AMD=1 WINO=1 STEPS=10 python3 examples/hlb_cifar10.py | tee train_cifar_wino.txt
    - name: Run full CIFAR training w 1 GPU
      run: time AMD=1 DEFAULT_FLOAT=HALF LATEWINO=1 STEPS=1000 TARGET_EVAL_ACC_PCT=93.2 python3 examples/hlb_cifar10.py | tee train_cifar_one_gpu.txt
    - name: Run full CIFAR training steps w 6 GPUS
@@ -473,7 +448,7 @@ jobs:
    - name: Run 10 MLPerf ResNet50 training steps (1 gpu)
      run: AMD=1 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=256 GPUS=1 MODEL=resnet python3 examples/mlperf/model_train.py | tee train_resnet_one_gpu.txt
    - name: Run 10 MLPerf ResNet50 training steps (6 gpu)
-      run: AMD=1 RUN_PROCESS_REPLAY=0 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=1536 GPUS=6 MODEL=resnet python3 examples/mlperf/model_train.py | tee train_resnet.txt
+      run: AMD=1 DEFAULT_FLOAT=HALF BENCHMARK=10 BS=1536 GPUS=6 MODEL=resnet python3 examples/mlperf/model_train.py | tee train_resnet.txt
    - uses: actions/upload-artifact@v4
      with:
        name: Speed (AMD Training)
@@ -493,7 +468,6 @@ jobs:
  testqualcommbenchmark:
    name: comma Benchmark
    runs-on: [self-hosted, Linux, comma]
-    timeout-minutes: 20
    defaults:
      run:
        shell: bash -o pipefail {0}
@@ -508,6 +482,10 @@ jobs:
        rm -f /tmp/staging.db /tmp/staging.db-shm /tmp/staging.db-wal
    - name: reset process replay
      run: test/external/process_replay/reset.py
+    - name: openpilot compile 0.9.4
+      run: PYTHONPATH=. NOLOCALS=1 FLOAT16=1 IMAGE=2 QCOM=1 taskset -c 4-7 python examples/openpilot/compile2.py | tee openpilot_compile_0_9_4.txt
+    - name: openpilot compile 0.9.7
+      run: PYTHONPATH=. NOLOCALS=1 FLOAT16=1 IMAGE=2 QCOM=1 taskset -c 4-7 python examples/openpilot/compile2.py https://github.com/commaai/openpilot/raw/v0.9.7/selfdrive/modeld/models/supercombo.onnx | tee openpilot_compile_0_9_7.txt
    - name: validate openpilot 0.9.7
      run: PYTHONPATH=. FLOAT16=0 IMAGE=2 QCOM=1 taskset -c 4-7 python3 test/external/external_benchmark_openpilot.py https://github.com/commaai/openpilot/raw/v0.9.7/selfdrive/modeld/models/supercombo.onnx | tee openpilot_image_0_9_7.txt
    - name: benchmark openpilot 0.9.4
@@ -1,7 +1,7 @@
 name: Unit Tests
 env:
  # increment this when downloads substantially change to avoid the internet
-  DOWNLOAD_CACHE_VERSION: '8'
+  DOWNLOAD_CACHE_VERSION: '6'
  RUN_PROCESS_REPLAY: 1
  GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
  PYTHONPATH: .
@@ -14,119 +14,6 @@ on:
  workflow_dispatch:

 jobs:
-  autogen:
-    name: Autogen+Docs
-    runs-on: ubuntu-latest
-    timeout-minutes: 20
-    steps:
-    - name: Checkout Code
-      uses: actions/checkout@v4
-    - name: Set up Python 3.12
-      uses: actions/setup-python@v5
-      with:
-        python-version: 3.12
-    - name: Install docs dependencies (no cache)
-      run: pip install -e '.[docs]'
-    - name: Use as an external package
-      run: |
-        mkdir $HOME/test_external_dir
-        cd $HOME/test_external_dir
-        python -m venv venv
-        source venv/bin/activate
-        pip install $GITHUB_WORKSPACE
-        python -c "from tinygrad.tensor import Tensor; print(Tensor([1,2,3,4,5]))"
-        pip install mypy
-        mypy -c "from tinygrad.tensor import Tensor; print(Tensor([1,2,3,4,5]))"
-    - name: Run beautiful_mnist with tinygrad only
-      run: |
-        mkdir $GITHUB_WORKSPACE/test_dir
-        cd $GITHUB_WORKSPACE/test_dir
-        python -m venv venv
-        source venv/bin/activate
-        pip install $GITHUB_WORKSPACE
-        cp $GITHUB_WORKSPACE/examples/beautiful_mnist.py .
-        PYTHONPATH=$GITHUB_WORKSPACE BS=2 STEPS=10 python beautiful_mnist.py
-    - name: Test Docs Build
-      run: python -m mkdocs build --strict
-    - name: Test Docs
-      run: |
-        python docs/abstractions2.py
-        python docs/abstractions3.py
-    - name: Test Quickstart
-      run: awk '/```python/{flag=1;next}/```/{flag=0}flag' docs/quickstart.md > quickstart.py &&  PYTHONPATH=. python quickstart.py
-    - name: Test DEBUG
-      run: DEBUG=100 python3 -c "from tinygrad import Tensor; N = 1024; a, b = Tensor.rand(N, N), Tensor.rand(N, N); c = (a.reshape(N, 1, N) * b.T.reshape(1, N, N)).sum(axis=2); print((c.numpy() - (a.numpy() @ b.numpy())).mean())"
-    - name: Install OpenCL
-      run: |
-        echo 'Acquire::http::Pipeline-Depth "5";' | sudo tee -a /etc/apt/apt.conf.d/99parallel
-        echo "deb [ allow-insecure=yes ] https://apt.repos.intel.com/oneapi all main" | sudo tee /etc/apt/sources.list.d/oneAPI.list
-        sudo apt update || true
-        sudo apt install --allow-unauthenticated -y --no-install-recommends opencl-headers \
-          intel-oneapi-runtime-openmp=2023.2.1-16 intel-oneapi-runtime-compilers-common=2023.2.1-16 intel-oneapi-runtime-compilers=2023.2.1-16 \
-          intel-oneapi-runtime-dpcpp-sycl-opencl-cpu=2023.2.1-16 intel-oneapi-runtime-tbb-common=2021.10.0-49541 \
-          intel-oneapi-runtime-tbb=2021.10.0-49541 intel-oneapi-runtime-opencl=2023.2.1-16
-    - name: Install packages (cuda)
-      run: |
-        echo 'Acquire::http::Pipeline-Depth "5";' | sudo tee -a /etc/apt/apt.conf.d/99parallel
-        sudo apt update -y || true
-        sudo apt install -y --no-install-recommends git g++ cmake ninja-build llvm-15-dev zlib1g-dev libglew-dev \
-          flex bison libfl-dev libboost-thread-dev libboost-filesystem-dev nvidia-cuda-toolkit-gcc libzstd-dev
-    - name: Install packages (amd)
-      run: |
-        echo 'Acquire::http::Pipeline-Depth "5";' | sudo tee -a /etc/apt/apt.conf.d/99parallel
-        wget https://repo.radeon.com/rocm/rocm.gpg.key -O - | gpg --dearmor | sudo tee /etc/apt/keyrings/rocm.gpg > /dev/null
-        sudo tee /etc/apt/sources.list.d/rocm.list <<'EOF'
-        deb [arch=amd64 signed-by=/etc/apt/keyrings/rocm.gpg] https://repo.radeon.com/rocm/apt/6.1.2 jammy main
-        EOF
-        echo -e 'Package: *\nPin: release o=repo.radeon.com\nPin-Priority: 600' | sudo tee /etc/apt/preferences.d/rocm-pin-600
-        sudo apt update || true
-        sudo apt install --no-install-recommends --allow-unauthenticated -y hsa-rocr comgr hsa-rocr-dev liburing-dev libc6-dev
-        curl -s https://api.github.com/repos/Qazalin/remu/releases/latest | \
-        jq -r '.assets[] | select(.name == "libremu.so").browser_download_url' | \
-        sudo xargs curl -L -o /usr/local/lib/libremu.so
-        sudo tee --append /etc/ld.so.conf.d/rocm.conf <<'EOF'
-          /opt/rocm/lib
-          /opt/rocm/lib64
-        EOF
-        sudo ldconfig
-    - name: Compile EfficientNet to C and test it
-      run: |
-        CLANG=1 PYTHONPATH="." python examples/compile_efficientnet.py > recognize.c
-        clang -O2 recognize.c -lm -o recognize
-        cat test/models/efficientnet/Chicken.jpg | ./recognize | grep cock
-    - name: Verify OpenCL autogen
-      run: |
-        cp tinygrad/runtime/autogen/opencl.py /tmp/opencl.py.bak
-        ./autogen_stubs.sh opencl
-        diff /tmp/opencl.py.bak tinygrad/runtime/autogen/opencl.py
-    - name: Verify CUDA autogen
-      run: |
-        cp tinygrad/runtime/autogen/cuda.py /tmp/cuda.py.bak
-        cp tinygrad/runtime/autogen/nv_gpu.py /tmp/nv_gpu.py.bak
-        ./autogen_stubs.sh cuda
-        ./autogen_stubs.sh nv
-        diff /tmp/cuda.py.bak tinygrad/runtime/autogen/cuda.py
-        diff /tmp/nv_gpu.py.bak tinygrad/runtime/autogen/nv_gpu.py
-    - name: Verify AMD autogen
-      run: |
-        cp tinygrad/runtime/autogen/hsa.py /tmp/hsa.py.bak
-        cp tinygrad/runtime/autogen/comgr.py /tmp/comgr.py.bak
-        cp tinygrad/runtime/autogen/amd_gpu.py /tmp/amd_gpu.py.bak
-        ./autogen_stubs.sh hsa
-        ./autogen_stubs.sh comgr
-        ./autogen_stubs.sh amd
-        diff /tmp/hsa.py.bak tinygrad/runtime/autogen/hsa.py
-        diff /tmp/comgr.py.bak tinygrad/runtime/autogen/comgr.py
-        diff /tmp/amd_gpu.py.bak tinygrad/runtime/autogen/amd_gpu.py
-    - name: Verify Linux autogen
-      run: |
-        cp tinygrad/runtime/autogen/libc.py /tmp/libc.py.bak
-        cp tinygrad/runtime/autogen/io_uring.py /tmp/io_uring.py.bak
-        ./autogen_stubs.sh libc
-        ./autogen_stubs.sh io_uring
-        diff /tmp/libc.py.bak tinygrad/runtime/autogen/libc.py
-        diff /tmp/io_uring.py.bak tinygrad/runtime/autogen/io_uring.py
-
  uops:
    name: uops tests
    runs-on: ubuntu-latest
@@ -150,9 +37,7 @@ jobs:
        IMAGE=2 PYTHON=1 python3 test/test_ops.py TestOps.test_gemm
        IMAGE=2 PYTHON=1 python3 test/test_ops.py TestOps.test_simple_conv2d
    - name: Test emulated METAL tensor cores
-      run: |
-        DEBUG=2 EMULATE_METAL=1 FORWARD_ONLY=1 PYTHON=1 python3 test/test_ops.py TestOps.test_big_gemm
-        PYTHONPATH=. DEBUG=2 EMULATE_METAL=1 FORWARD_ONLY=1 PYTHON=1 python3 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded
+      run: DEBUG=2 EMULATE_METAL=1 FORWARD_ONLY=1 PYTHON=1 python3 test/test_ops.py TestOps.test_big_gemm
    - name: Test emulated AMX tensor cores
      run: PYTHONPATH=. DEBUG=2 AMX=1 EMULATE_AMX=1 FORWARD_ONLY=1 PYTHON=1 python3 test/test_ops.py TestOps.test_gemm
    - name: Test emulated AMD tensor cores
@@ -161,11 +46,8 @@ jobs:
        PYTHONPATH=. DEBUG=2 EMULATE_AMD=1 FORWARD_ONLY=1 PYTHON=1 N=64 HALF=1 ACC_HALF=0 python3 ./extra/gemm/simple_matmul.py
        PYTHONPATH=. DEBUG=2 EMULATE_AMD=1 FORWARD_ONLY=1 PYTHON=1 N=16 HALF=1 ACC_HALF=1 python3 ./extra/gemm/simple_matmul.py
        PYTHONPATH=. DEBUG=2 EMULATE_AMD=1 FORWARD_ONLY=1 PYTHON=1 N=64 HALF=1 ACC_HALF=1 python3 ./extra/gemm/simple_matmul.py
-        PYTHONPATH=. DEBUG=2 EMULATE_AMD=1 FORWARD_ONLY=1 PYTHON=1 python3 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded
    - name: Test emulated CUDA tensor cores
-      run: |
-        DEBUG=2 EMULATE_CUDA=1 FORWARD_ONLY=1 PYTHON=1 python3 test/test_ops.py TestOps.test_gemm_fp16
-        PYTHONPATH="." DEBUG=2 EMULATE_CUDA=1 FORWARD_ONLY=1 PYTHON=1 python3 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded
+      run: DEBUG=2 EMULATE_CUDA=1 FORWARD_ONLY=1 PYTHON=1 python3 test/test_ops.py TestOps.test_gemm_fp16
    - name: Test emulated INTEL OpenCL tensor cores
      run: DEBUG=2 EMULATE_INTEL=1 FORWARD_ONLY=1 PYTHON=1 HALF=1 N=64 python3 ./extra/gemm/simple_matmul.py
    - name: Full test tensor cores
@@ -192,7 +74,7 @@ jobs:
    - name: Test dtype with Python emulator
      run: DEBUG=1 PYTHONPATH=. PYTHON=1 python3 -m pytest -n=auto test/test_dtype.py test/test_dtype_alu.py
    - name: Test ops with Python emulator
-      run: DEBUG=2 PYTHON=1 python3 -m pytest -n=auto test/test_ops.py -k "not (test_split or test_simple_cumsum or test_cumsum or test_einsum or test_dot or test_dot_1d or test_big_gemm or test_broadcastdot or test_multidot or test_var_axis or test_std_axis or test_broadcast_full or test_broadcast_partial or test_simple_conv3d or test_dilated_conv_transpose2d or test_simple_conv_transpose3d or test_large_input_conv2d or test_max_pool2d or test_max_pool2d_simple or test_max_pool2d_bigger_stride or test_avg_pool2d or test_cat or test_scaled_product_attention or test_scaled_product_attention_causal or test_slice_fancy_indexing_dim_inject_none or test_slice_fancy_indexing_list_indices or test_slice_fancy_indexing_no_dim_collapse or test_slice_fancy_indexing_tuple_indices or test_slice_fancy_indexing_list_with_tensors or test_slice_fancy_indexing_dim_collapse_int or test_interpolate_bilinear or test_interpolate_bilinear_corners_aligned)" --durations=20
+      run: DEBUG=2 PYTHON=1 python3 -m pytest -n=auto test/test_ops.py -k "not (test_split or test_simple_cumsum or test_cumsum or test_einsum or test_dot or test_dot_1d or test_big_gemm or test_broadcastdot or test_multidot or test_var_axis or test_std_axis or test_broadcast_full or test_broadcast_partial or test_simple_conv3d or test_dilated_conv_transpose2d or test_simple_conv_transpose3d or test_large_input_conv2d or test_maxpool2d or test_maxpool2d_simple or test_maxpool2d_bigger_stride or test_avgpool2d or test_cat or test_scaled_product_attention or test_scaled_product_attention_causal or test_slice_fancy_indexing_dim_inject_none or test_slice_fancy_indexing_list_indices or test_slice_fancy_indexing_no_dim_collapse or test_slice_fancy_indexing_tuple_indices or test_slice_fancy_indexing_list_with_tensors or test_slice_fancy_indexing_dim_collapse_int or test_interpolate_bilinear or test_interpolate_bilinear_corners_aligned)" --durations=20
    - name: Test uops with Python emulator
      run: PYTHON=1 python3 -m pytest test/test_uops.py --durations=20
    - name: Test symbolic with Python emulator
@@ -201,7 +83,7 @@ jobs:
      run: PYTHONPATH=. PYTHON=1 python3 -m pytest -rA test/test_linearizer_failures.py::TestLinearizerFailures::test_failure_1

  linter:
-    name: Linters+fuzz+unit Tests
+    name: Linters
    runs-on: ubuntu-latest
    timeout-minutes: 20

@@ -209,17 +91,17 @@ jobs:
    steps:
    - name: Checkout Code
      uses: actions/checkout@v4
-    - name: Set up Python 3.10
+    - name: Set up Python 3.8
      uses: actions/setup-python@v5
      with:
-        python-version: "3.10"
+        python-version: 3.8
    - name: Cache python packages
      uses: actions/cache@v4
      with:
-        path: ${{ env.Python3_ROOT_DIR }}/lib/python3.10/site-packages
-        key: linting-packages-${{ hashFiles('**/setup.py') }}-3.10
+        path: ${{ env.Python3_ROOT_DIR }}/lib/python3.8/site-packages
+        key: linting-packages-${{ hashFiles('**/setup.py') }}-3.8
    - name: Install dependencies
-      run: pip install -e '.[linting,testing]' --extra-index-url https://download.pytorch.org/whl/cpu
+      run: pip install -e '.[linting,testing,docs]' --extra-index-url https://download.pytorch.org/whl/cpu
    - name: Lint bad-indentation and trailing-whitespace with pylint
      run: python -m pylint --disable=all -e W0311 -e C0303 --jobs=0 --indent-string='  ' --recursive=y .
    - name: Lint with ruff
@@ -229,7 +111,15 @@ jobs:
    - name: Lint tinygrad with pylint
      run: python -m pylint tinygrad/
    - name: Run mypy
-      run: python -m mypy --strict-equality --lineprecision-report . && cat lineprecision.txt
+      run: python -m mypy --strict-equality
+    - name: Test Docs
+      run: |
+        python docs/abstractions2.py
+        python docs/abstractions3.py
+    - name: Test Docs Build
+      run: python -m mkdocs build --strict
+    - name: Test Quickstart
+      run: awk '/```python/{flag=1;next}/```/{flag=0}flag' docs/quickstart.md > quickstart.py &&  PYTHONPATH=. python quickstart.py
    - name: Test README
      run: awk '/```python/{flag=1;next}/```/{flag=0}flag' README.md > README.py &&  PYTHONPATH=. python README.py
    - name: Run unit tests
@@ -240,6 +130,27 @@ jobs:
      run: |
        PYTHONPATH="." python test/external/fuzz_shapetracker.py
        PYTHONPATH="." python test/external/fuzz_shapetracker_math.py
+    - name: Use as an external package
+      run: |
+        mkdir $HOME/test_external_dir
+        cd $HOME/test_external_dir
+        python -m venv venv
+        source venv/bin/activate
+        pip install $GITHUB_WORKSPACE
+        python -c "from tinygrad.tensor import Tensor; print(Tensor([1,2,3,4,5]))"
+        pip install mypy
+        mypy -c "from tinygrad.tensor import Tensor; print(Tensor([1,2,3,4,5]))"
+    - name: Run beautiful_mnist without numpy
+      run: |
+        mkdir $HOME/test_no_numpy_dir
+        cd $HOME/test_no_numpy_dir
+        python -m venv venv
+        source venv/bin/activate
+        pip install $GITHUB_WORKSPACE
+        cp $GITHUB_WORKSPACE/examples/beautiful_mnist.py .
+        PYTHONPATH=$GITHUB_WORKSPACE BS=2 STEPS=10 python beautiful_mnist.py
+    - name: Test DEBUG
+      run: DEBUG=100 python3 -c "from tinygrad import Tensor; N = 1024; a, b = Tensor.rand(N, N), Tensor.rand(N, N); c = (a.reshape(N, 1, N) * b.T.reshape(1, N, N)).sum(axis=2); print((c.numpy() - (a.numpy() @ b.numpy())).mean())"
    - name: Repo line count <= 9999 lines
      run: MAX_LINE_COUNT=9999 python sz.py

@@ -290,18 +201,28 @@ jobs:
      - if: ${{ matrix.task == 'optimage'}}
        name: Test GPU IMAGE=2 ops + training
        run: |
-          PYTHONPATH="." GPU=1 IMAGE=2 python -m pytest -n=auto test/test_ops.py --durations=20
+          PYTHONPATH="." GPU=1 IMAGE=2 python -m pytest -n=auto test/test_ops.py
          PYTHONPATH="." GPU=1 IMAGE=2 python3 test/models/test_end2end.py TestEnd2End.test_linear_mnist
      - if: ${{ matrix.task == 'optimage' }}
-        name: Test openpilot model kernel count and gate usage
+        name: Test openpilot model compile and size
        run: |
-          PYTHONPATH="." ALLOWED_KERNEL_COUNT=208 ALLOWED_READ_IMAGE=2131 ALLOWED_GATED_READ_IMAGE=13 FLOAT16=0 GPU=1 IMAGE=2 python examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/v0.9.4/selfdrive/modeld/models/supercombo.onnx
+          PYTHONPATH="." DEBUG=2 ALLOWED_KERNEL_COUNT=208 ALLOWED_GATED_READ_IMAGE=13 FLOAT16=1 DEBUGCL=1 GPU=1 IMAGE=2 python examples/openpilot/compile2.py
+          python -c 'import os; assert os.path.getsize("/tmp/output.thneed") < 100_000_000'
+      - if: ${{ matrix.task == 'optimage' }}
+        name: Test openpilot model correctness (float32)
+        run: PYTHONPATH="." FLOAT16=0 DEBUGCL=1 GPU=1 IMAGE=2 python examples/openpilot/compile2.py
      - if: ${{ matrix.task == 'optimage' }}
        name: Test openpilot alt model correctness (float32)
-        run: PYTHONPATH="." FLOAT16=0 DEBUGCL=1 GPU=1 IMAGE=2 python examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/3799fe46b3a629e491d4b8498b8ae83e4c88c304/selfdrive/modeld/models/supercombo.onnx
+        run: PYTHONPATH="." FLOAT16=0 DEBUGCL=1 GPU=1 IMAGE=2 python examples/openpilot/compile2.py https://github.com/commaai/openpilot/raw/3799fe46b3a629e491d4b8498b8ae83e4c88c304/selfdrive/modeld/models/supercombo.onnx
      - if: ${{ matrix.task == 'optimage' }}
        name: Test openpilot fastvits model correctness (float32)
-        run: PYTHONPATH="." FLOAT16=0 DEBUGCL=1 GPU=1 IMAGE=2 python examples/openpilot/compile3.py https://github.com/commaai/openpilot/raw/9118973ed03c1ae1d40cf69a29507ec2cc78efd7/selfdrive/modeld/models/supercombo.onnx
+        run: PYTHONPATH="." FLOAT16=0 DEBUGCL=1 GPU=1 IMAGE=2 python examples/openpilot/compile2.py https://github.com/commaai/openpilot/raw/9118973ed03c1ae1d40cf69a29507ec2cc78efd7/selfdrive/modeld/models/supercombo.onnx
+      - if: ${{ matrix.task == 'optimage' }}
+        name: Compile EfficientNet to C and test it
+        run: |
+          CLANG=1 PYTHONPATH="." python examples/compile_efficientnet.py > recognize.c
+          clang -O2 recognize.c -lm -o recognize
+          cat test/models/efficientnet/Chicken.jpg | ./recognize | grep cock
      - if: ${{ matrix.task == 'onnx' }}
        name: Test ONNX (GPU)
        run: GPU=1 python -m pytest -n=auto test/external/external_test_onnx_backend.py --durations=20
@@ -310,16 +231,10 @@ jobs:
        run: CLANG=1 python -m pytest -n=auto test/external/external_test_onnx_backend.py --durations=20
      - if: ${{ matrix.task == 'onnx' }}
        name: Run CLOUD=1 Test
-        run: |
-          CLOUDDEV=CLANG CLOUD=1 python3 test/test_tiny.py
-          CLOUDDEV=GPU CLOUD=1 python3 test/test_tiny.py
-          CLOUDDEV=GPU IMAGE=2 CLOUD=1 python3 test/test_tiny.py
-      - if: ${{ matrix.task == 'onnx' }}
-        name: Test Optimization Helpers
-        run: PYTHONPATH="." DEBUG=1 python3 extra/optimization/test_helpers.py
+        run: CLOUDDEV=CLANG CLOUD=1 python3 test/test_ops.py TestOps.test_tiny_add
      - if: ${{ matrix.task == 'onnx' }}
        name: Test Action Space
-        run: PYTHONPATH="." DEBUG=1 GPU=1 python3 extra/optimization/get_action_space.py
+        run: PYTHONPATH="." GPU=1 python3 extra/optimization/get_action_space.py
      - if: ${{ matrix.task == 'onnx' }}
        name: Test Beam Search
        run: PYTHONPATH="." GPU=1 IGNORE_BEAM_CACHE=1 python3 -m pytest extra/optimization/test_beam_search.py
@@ -344,47 +259,46 @@ jobs:
          export COMMIT_MESSAGE=$(git show -s --format=%B ${{ github.event.pull_request.head.sha }})
          cp test/external/process_replay/process_replay.py ./process_replay.py && git fetch origin master && git -c advice.detachedHead=false checkout origin/master && PYTHONPATH=. python3 process_replay.py

-  testwebgpu:
-    name: WebGPU Tests
-    runs-on: ubuntu-latest
-    timeout-minutes: 20
-    steps:
-    - name: Checkout Code
-      uses: actions/checkout@v4
-    - name: Set up Python 3.11
-      uses: actions/setup-python@v5
-      with:
-        python-version: 3.11
-    - name: Cache python packages
-      uses: actions/cache@v4
-      with:
-        path: ~/.local/lib/python3.11/site-packages
-        key: webgpu-testing-user3-packages-${{ hashFiles('**/setup.py') }}
-    - name: Install Dependencies
-      run: pip install --user -e '.[webgpu,testing]' --extra-index-url https://download.pytorch.org/whl/cpu
-    - name: Install dependencies for software-based vulkan
-      run: |
-        sudo apt update -y || true
-        sudo apt install -y libegl1-mesa libgl1-mesa-dri libxcb-xfixes0-dev mesa-vulkan-drivers
-    - name: Cache downloads
-      uses: actions/cache@v4
-      with:
-        path: ~/.cache/tinygrad/downloads/
-        key: downloads-cache-webgpu-${{ env.DOWNLOAD_CACHE_VERSION }}
-    - name: Check Device.DEFAULT (WEBGPU) and print some source
-      run: |
-        WEBGPU=1 python -c "from tinygrad import Device; assert Device.DEFAULT == 'WEBGPU', Device.DEFAULT"
-        WEBGPU=1 DEBUG=4 FORWARD_ONLY=1 python3 test/test_ops.py TestOps.test_add
-    - name: Run selected webgpu tests
-      run: |
-          WEBGPU=1 WGPU_BACKEND_TYPE=Vulkan python3 -m pytest -n=auto test/ --ignore=test/external --ignore=test/models --ignore=test/unit \
-          --ignore=test/test_copy_speed.py --ignore=test/test_rearrange_einops.py --ignore=test/test_speed_v_torch.py --ignore=test/test_transcendental.py \
-          --ignore=test/test_fuzz_shape_ops.py --ignore=test/test_linearizer_failures.py --durations=20
-    - name: Run process replay tests
-      run: |
-        export PR_TITLE=$(jq -r .pull_request.title "$GITHUB_EVENT_PATH")
-        export COMMIT_MESSAGE=$(git show -s --format=%B ${{ github.event.pull_request.head.sha }})
-        cp test/external/process_replay/process_replay.py ./process_replay.py && git fetch origin master && git -c advice.detachedHead=false checkout origin/master && PYTHONPATH=. python3 process_replay.py
+  #testwebgpu:
+  #  name: WebGPU Tests
+  #  runs-on: macos-13
+  #  timeout-minutes: 20
+  #  steps:
+  #  - name: Checkout Code
+  #    uses: actions/checkout@v4
+  #  - name: Set up Python 3.11
+  #    uses: actions/setup-python@v5
+  #    with:
+  #      python-version: 3.11
+  #  - name: Cache python packages
+  #    uses: actions/cache@v4
+  #    with:
+  #      path: /Users/runner/Library/Python/3.11/lib/python/site-packages
+  #      key: webgpu-testing-user3-packages-${{ hashFiles('**/setup.py') }}
+  #  - name: Install Dependencies
+  #    run: pip install --user -e '.[webgpu,testing]' --extra-index-url https://download.pytorch.org/whl/cpu
+  #  - name: Cache downloads
+  #    uses: actions/cache@v4
+  #    with:
+  #      path: ~/Library/Caches/tinygrad/downloads/
+  #      key: downloads-cache-webgpu-${{ env.DOWNLOAD_CACHE_VERSION }}
+  #  - name: Check Device.DEFAULT (WEBGPU) and print some source
+  #    run: |
+  #      WEBGPU=1 python -c "from tinygrad import Device; assert Device.DEFAULT == 'WEBGPU', Device.DEFAULT"
+  #      WEBGPU=1 DEBUG=4 FORWARD_ONLY=1 python3 test/test_ops.py TestOps.test_add
+    #- name: Run webgpu pytest
+    #  run: WEBGPU=1 WGPU_BACKEND_TYPE=Metal python -m pytest -n=auto
+  #  - name: Run selected webgpu tests
+  #    run: |
+  #      WEBGPU=1 WGPU_BACKEND_TYPE=Metal python -m pytest -n=auto test/test_ops.py test/test_dtype.py \
+  #      test/test_jit.py test/test_symbolic_ops.py test/test_symbolic_jit.py test/test_linearizer.py \
+  #      test/test_linearizer_failures.py test/test_nn.py
+  #  - name: Build WEBGPU Efficientnet
+  #    run: WEBGPU=1 WGPU_BACKEND_TYPE=Metal python -m examples.compile_efficientnet
+  #  - name: Install Puppeteer
+  #    run: npm install puppeteer
+  #  - name: Run WEBGPU Efficientnet
+  #    run: node test/web/test_webgpu.js

  testmetal:
    name: Metal Tests
@@ -406,7 +320,7 @@ jobs:
        path: /Users/runner/Library/Python/3.11/lib/python/site-packages
        key: metal-m1-testing-user3-packages-${{ hashFiles('**/setup.py') }}
    - name: Install Dependencies
-      run: pip install --user -e '.[webgpu,testing]' --extra-index-url https://download.pytorch.org/whl/cpu
+      run: pip install --user -e '.[testing]' --extra-index-url https://download.pytorch.org/whl/cpu
    - name: Cache downloads
      uses: actions/cache@v4
      with:
@@ -431,26 +345,50 @@ jobs:
    - name: Test Beam Search
      run: PYTHONPATH="." METAL=1 IGNORE_BEAM_CACHE=1 python3 -m pytest extra/optimization/test_beam_search.py
    - name: Fuzz Test linearizer
-      run: PYTHONPATH="." METAL=1 DEPTH=4 FUZZ_N=50 FUZZ_MAX_SIZE=1000000 python test/external/fuzz_linearizer.py
-    # - name: Fuzz Test models schedule
-    # run: FUZZ_SCHEDULE=1 FUZZ_SCHEDULE_MAX_PATHS=5 python -m pytest test/models/test_train.py test/models/test_end2end.py
+      run: PYTHONPATH="." METAL=1 FUZZ_ALL_ACTIONS=1 DEPTH=2 FUZZ_N=24 FUZZ_MAX_SIZE=10000000 python test/external/fuzz_linearizer.py
+    - name: Fuzz Test models schedule
+      run: FUZZ_SCHEDULE=1 FUZZ_SCHEDULE_MAX_PATHS=5 python -m pytest test/models/test_train.py test/models/test_end2end.py
    - name: Run TRANSCENDENTAL math
      run: TRANSCENDENTAL=2 python -m pytest -n=auto test/test_ops.py::TestOps::test_sin test/test_ops.py::TestOps::test_cos test/test_ops.py::TestOps::test_tan test/test_ops.py::TestOps::test_exp test/test_ops.py::TestOps::test_log --durations=20
-    # WebGPU e2e tests
-    - name: Build WEBGPU Efficientnet
-      run: WEBGPU=1 WGPU_BACKEND_TYPE=Metal python3 -m examples.compile_efficientnet
-    - name: Clean npm cache
-      run: npm cache clean --force
-    - name: Install Puppeteer
-      run: npm install puppeteer
-    - name: Run WEBGPU Efficientnet
-      run: node test/web/test_webgpu.js
    - name: Run process replay tests
      run: |
        export PR_TITLE=$(jq -r .pull_request.title "$GITHUB_EVENT_PATH")
        export COMMIT_MESSAGE=$(git show -s --format=%B ${{ github.event.pull_request.head.sha }})
        cp test/external/process_replay/process_replay.py ./process_replay.py && git fetch origin master && git -c advice.detachedHead=false checkout origin/master && PYTHONPATH=. python3 process_replay.py

+#  testwebgl:
+#    name: WebGL Tests
+#    runs-on: ubuntu-latest
+#    timeout-minutes: 20
+#
+#    steps:
+#    - name: Checkout Code
+#      uses: actions/checkout@v3
+#    - name: Set up Python 3.11
+#      uses: actions/setup-python@v4
+#      with:
+#        python-version: 3.11
+#    - name: Cache python packages
+#      uses: actions/cache@v4
+#      with:
+#        path: ${{ env.Python3_ROOT_DIR }}/lib/python3.11/site-packages
+#        key: webgl-testing-packages-${{ hashFiles('**/setup.py') }}
+#    - name: Install Dependencies
+#      run: pip install -e '.[webgl,testing]' --extra-index-url https://download.pytorch.org/whl/cpu
+#    - name: Cache downloads
+#      uses: actions/cache@v4
+#      with:
+#        path: ~/Library/Caches/tinygrad/downloads/
+#        key: downloads-cache-webgl-${{ env.DOWNLOAD_CACHE_VERSION }}
+#    - name: Prepare
+#      run: |
+#        sudo apt-get -y install xvfb
+#        sudo /usr/bin/Xvfb :0 -screen 0 4096x4096x24+32 &
+#    - name: Run selected webgl tests
+#      run: WEBGL=1 python -m pytest -n=auto test/test_ops.py test/test_dtype.py test/test_jit.py
+#    - name: Build WebGL Efficientnet
+#      run: WEBGL=1 python -m examples.compile_efficientnet
+
  tests:
    strategy:
      fail-fast: false
@@ -458,7 +396,7 @@ jobs:
        backend: [llvm, clang, gpu, ptx, amd, nv] #, triton]

    name: Tests on (${{ matrix.backend }})
-    runs-on: ubuntu-22.04
+    runs-on: ubuntu-latest
    timeout-minutes: 20

    steps:
@@ -466,14 +404,14 @@ jobs:
        uses: actions/checkout@v4
        with:
          fetch-depth: 2 # NOTE: this fetches the HEAD commit of the PR
-      - name: Set up Python 3.12
+      - name: Set up Python 3.11
        uses: actions/setup-python@v5
        with:
-          python-version: 3.12
+          python-version: 3.11
      - name: Cache python packages
        uses: actions/cache@v4
        with:
-          path: ${{ env.Python3_ROOT_DIR }}/lib/python3.12/site-packages
+          path: ${{ env.Python3_ROOT_DIR }}/lib/python3.11/site-packages
          key: ${{ matrix.backend }}-packages-${{ hashFiles('**/setup.py') }}
      - name: Cache downloads
        uses: actions/cache@v4
@@ -507,7 +445,7 @@ jobs:
          cache-name: cache-gpuocelot-build
        with:
          path: ${{ github.workspace }}/gpuocelot/ocelot
-          key: ubuntu22.04-gpuocelot-4524e34adb7eaccc6f71262f2e21d7052bb17c2f-rebuild-9
+          key: ubuntu22.04-gpuocelot-4524e34adb7eaccc6f71262f2e21d7052bb17c2f-rebuild-8
      - name: Clone/compile gpuocelot
        if: (matrix.backend == 'ptx' || matrix.backend == 'triton' || matrix.backend == 'nv') && steps.cache-build.outputs.cache-hit != 'true'
        run: |
@@ -522,7 +460,7 @@ jobs:
        if: matrix.backend == 'ptx' || matrix.backend == 'triton' || matrix.backend == 'nv'
        run: |
          cd ${{ github.workspace }}/gpuocelot/ocelot/build
-          sudo cp libgpuocelot.so /usr/lib/libgpuocelot.so
+          sudo ninja install -d explain
      - name: Install packages (amd)
        if: matrix.backend == 'amd'
        run: |
@@ -548,9 +486,48 @@ jobs:
        run: |
          PYTHONPATH=${{ github.workspace }} python3 -c "from tinygrad import Device; assert Device.DEFAULT in ['LLVM','CLANG','CUDA','GPU','AMD','NV'], Device.DEFAULT"
          DEBUG=5 PYTHONPATH=${{ github.workspace }} FORWARD_ONLY=1 python3 test/test_ops.py TestOps.test_add
+      - name: Verify OpenCL autogen
+        if: matrix.backend == 'gpu'
+        run: |
+          cp tinygrad/runtime/autogen/opencl.py /tmp/opencl.py.bak
+          ./autogen_stubs.sh opencl
+          diff /tmp/opencl.py.bak tinygrad/runtime/autogen/opencl.py
+      - name: Verify CUDA autogen
+        if: matrix.backend == 'nv'
+        run: |
+          cp tinygrad/runtime/autogen/cuda.py /tmp/cuda.py.bak
+          cp tinygrad/runtime/autogen/nv_gpu.py /tmp/nv_gpu.py.bak
+          ./autogen_stubs.sh cuda
+          ./autogen_stubs.sh nv
+          diff /tmp/cuda.py.bak tinygrad/runtime/autogen/cuda.py
+          diff /tmp/nv_gpu.py.bak tinygrad/runtime/autogen/nv_gpu.py
+      - name: Verify AMD autogen
+        if: matrix.backend == 'amd'
+        run: |
+          cp tinygrad/runtime/autogen/hsa.py /tmp/hsa.py.bak
+          cp tinygrad/runtime/autogen/comgr.py /tmp/comgr.py.bak
+          cp tinygrad/runtime/autogen/amd_gpu.py /tmp/amd_gpu.py.bak
+          ./autogen_stubs.sh hsa
+          ./autogen_stubs.sh comgr
+          ./autogen_stubs.sh amd
+          diff /tmp/hsa.py.bak tinygrad/runtime/autogen/hsa.py
+          diff /tmp/comgr.py.bak tinygrad/runtime/autogen/comgr.py
+          diff /tmp/amd_gpu.py.bak tinygrad/runtime/autogen/amd_gpu.py
+      - name: Verify Linux autogen
+        if: matrix.backend == 'amd'
+        run: |
+          cp tinygrad/runtime/autogen/libc.py /tmp/libc.py.bak
+          cp tinygrad/runtime/autogen/io_uring.py /tmp/io_uring.py.bak
+          ./autogen_stubs.sh libc
+          ./autogen_stubs.sh io_uring
+          diff /tmp/libc.py.bak tinygrad/runtime/autogen/libc.py
+          diff /tmp/io_uring.py.bak tinygrad/runtime/autogen/io_uring.py
      - name: Run pytest (not cuda or amd)
        if: matrix.backend!='ptx' && matrix.backend!='triton' && matrix.backend != 'amd' && matrix.backend != 'nv'
        run: python -m pytest -n=auto test/ --ignore=test/unit --durations=20
+      # - name: Run test_ops with FUZZ_UOPS=1
+      #   if: matrix.backend!='cuda' && matrix.backend!='ptx' && matrix.backend!='triton' && matrix.backend != 'amd' && matrix.backend != 'nv'
+      #   run: FUZZ_UOPS=1 python -m pytest -n=auto test/test_ops.py --durations=20
      - name: Run ONNX (only LLVM)
        if: matrix.backend == 'llvm'
        run: python -m pytest -n=auto test/external/external_test_onnx_backend.py --durations=20
@@ -35,9 +35,8 @@ extra/datasets/COCO/
 extra/datasets/audio*
 extra/weights
 venv
-examples/**/**/net*.*[js,json]
+examples/**/net.*[js,json]
 examples/**/*.safetensors
-examples/webgpu/stable_diffusion/*.*[js,mjs]
 node_modules
 package.json
 package-lock.json
@@ -1,28 +1,33 @@
-# on Windows -- $env:SKIP="tests,example"
 repos:
  - repo: local
    hooks:
-      - id: ruff
-        name: ruff
-        entry: python3 -m ruff check .
+      - id: whitespace
+        name: strip whitespace
+        entry: find tinygrad -type f -name "*.py" -exec sed -i '' 's/ *$//' '{}' ';'
        language: system
        always_run: true
        pass_filenames: false
-      - id: tiny
-        name: tiny tests
-        entry: python3 -m pytest test/test_tiny.py
+      - id: ruff
+        name: ruff
+        entry: ruff check .
        language: system
        always_run: true
        pass_filenames: false
      - id: mypy
        name: mypy
-        entry: python3 -m mypy tinygrad/ --strict-equality
+        entry: mypy tinygrad/ --strict-equality
+        language: system
+        always_run: true
+        pass_filenames: false
+      - id: docs2
+        name: docs2
+        entry: python3 docs/abstractions2.py
        language: system
        always_run: true
        pass_filenames: false
      - id: devicetests
        name: select GPU tests
-        entry: env GPU=1 PYTHONPATH="." python3 -m pytest test/test_uops.py test/test_search.py
+        entry: env GPU=1 PYTHONPATH="." pytest test/test_uops.py test/test_search.py
        language: system
        always_run: true
        pass_filenames: false
@@ -40,7 +45,7 @@ repos:
        pass_filenames: false
      - id: pylint
        name: pylint
-        entry: python3 -m pylint tinygrad/
+        entry: env PYTHONPATH="." python3 -m pylint tinygrad/
        language: system
        always_run: true
        pass_filenames: false
@@ -54,7 +54,7 @@ confidence=
 # --enable=similarities". If you want to run only the classes checker, but have
 # no Warning level messages displayed, use"--disable=all --enable=classes
 # --disable=W"
-disable=C,R,W0613,W0511,W0212,W0201,W0106,W0603,W0621,W0703,W1201,W1203,E1136,W1514,E1101,W0221,W0105,E0401,abstract-method
+disable=C,R,W0613,W0511,W0212,W0201,W0106,W0603,W0621,W0703,W1201,W1203,E1136,W1514,E1101,W0221,W0105,E0401
 # E1101 for function binding
 # W0221 for Function class
 # W0105 for comment strings
@@ -88,7 +88,6 @@ tinygrad already supports numerous accelerators, including:
 - [x] [AMD](tinygrad/runtime/ops_amd.py)
 - [x] [NV](tinygrad/runtime/ops_nv.py)
 - [x] [QCOM](tinygrad/runtime/ops_qcom.py)
- [x] [WEBGPU](tinygrad/runtime/ops_webgpu.py)

 And it is easy to add more! Your accelerator of choice only needs to support a total of ~25 low level ops.

@@ -126,8 +125,8 @@ y = Tensor([[2.0,0,-2.0]], requires_grad=True)
 z = y.matmul(x).sum()
 z.backward()

-print(x.grad.tolist())  # dz/dx
-print(y.grad.tolist())  # dz/dy
+print(x.grad.numpy())  # dz/dx
+print(y.grad.numpy())  # dz/dy
 ```

 The same thing but in PyTorch:
@@ -139,8 +138,8 @@ y = torch.tensor([[2.0,0,-2.0]], requires_grad=True)
 z = y.matmul(x).sum()
 z.backward()

-print(x.grad.tolist())  # dz/dx
-print(y.grad.tolist())  # dz/dy
+print(x.grad.numpy())  # dz/dx
+print(y.grad.numpy())  # dz/dy
 ```

 ## Contributing
@@ -179,4 +178,4 @@ python3 -m pytest test/                 # whole test suite

 #### Process replay tests

-[Process replay](https://github.com/tinygrad/tinygrad/blob/master/test/external/process_replay/README.md) compares your PR's generated kernels against master. If your PR is a refactor or speedup without any expected behavior change, It should include [pr] in the pull request title.
+[Process replay](https://github.com/tinygrad/tinygrad/blob/master/test/external/process_replay/process_replay.py) compares your PR's generated kernels against master. If your PR is a refactor or speedup without any expected behavior change, It should include [pr] in the pull request title, [example](https://github.com/tinygrad/tinygrad/pull/4995). Note that you should keep your branch up-to-date with master.
@@ -117,13 +117,11 @@ generate_nv() {
    $NVKERN_SRC/src/common/sdk/nvidia/inc/class/clc56f.h \
    $NVKERN_SRC/src/common/sdk/nvidia/inc/class/clc56f.h \
    $NVKERN_SRC/src/common/sdk/nvidia/inc/class/clc56f.h \
-    $NVKERN_SRC/src/common/sdk/nvidia/inc/class/cl83de.h \
    $NVKERN_SRC/src/nvidia/generated/g_allclasses.h \
    $NVKERN_SRC/src/common/sdk/nvidia/inc/class/clc6c0.h \
    $NVKERN_SRC/kernel-open/nvidia-uvm/clc6b5.h \
    $NVKERN_SRC/kernel-open/nvidia-uvm/uvm_ioctl.h \
    $NVKERN_SRC/kernel-open/nvidia-uvm/uvm_linux_ioctl.h \
-    $NVKERN_SRC/kernel-open/nvidia-uvm/hwref/ampere/ga100/dev_fault.h \
    $NVKERN_SRC/src/nvidia/arch/nvalloc/unix/include/nv_escape.h \
    $NVKERN_SRC/src/nvidia/arch/nvalloc/unix/include/nv-ioctl.h \
    $NVKERN_SRC/src/nvidia/arch/nvalloc/unix/include/nv-ioctl-numbers.h \
@@ -162,8 +160,6 @@ generate_amd() {
  clang2py -k cdefstum \
    extra/hip_gpu_driver/sdma_registers.h \
    extra/hip_gpu_driver/nvd.h \
-    extra/hip_gpu_driver/kfd_pm4_headers_ai.h \
-    extra/hip_gpu_driver/soc21_enum.h \
    extra/hip_gpu_driver/sdma_v6_0_0_pkt_open.h \
    extra/hip_gpu_driver/gc_11_0_0_offset.h \
    extra/hip_gpu_driver/gc_10_3_0_offset.h \
@@ -213,7 +209,7 @@ generate_libc() {

  sed -i "s\import ctypes\import ctypes, ctypes.util, os\g" $BASE/libc.py
  sed -i "s\FIXME_STUB\libc\g" $BASE/libc.py
-  sed -i "s\FunctionFactoryStub()\None if (libc_path := ctypes.util.find_library('c')) is None else ctypes.CDLL(libc_path)\g" $BASE/libc.py
+  sed -i "s\FunctionFactoryStub()\ctypes.CDLL(ctypes.util.find_library('c'))\g" $BASE/libc.py

  fixup $BASE/libc.py
 }
@@ -15,8 +15,8 @@ a = MallocAllocator.alloc(4)
 b = MallocAllocator.alloc(4)

 # load in some values (little endian)
-MallocAllocator._copyin(a, memoryview(bytearray([2,0,0,0])))
-MallocAllocator._copyin(b, memoryview(bytearray([3,0,0,0])))
+MallocAllocator.copyin(a, bytearray([2,0,0,0]))
+MallocAllocator.copyin(b, bytearray([3,0,0,0]))

 # compile a program to a binary
 lib = ClangCompiler().compile("void add(int *out, int *a, int *b) { out[0] = a[0] + b[0]; }")
@@ -28,7 +28,7 @@ fxn = ClangProgram("add", lib)
 fxn(out, a, b)

 # check the data out
-print(val := MallocAllocator._as_buffer(out).cast("I").tolist()[0])
+print(val := MallocAllocator.as_buffer(out).cast("I").tolist()[0])
 assert val == 5


@@ -37,9 +37,9 @@ print("******** second, the Device ***********")
 DEVICE = "CLANG"   # NOTE: you can change this!

 import struct
-from tinygrad.dtype import dtypes
+from tinygrad.dtype import PtrDType, dtypes
 from tinygrad.device import Buffer, Device
-from tinygrad.ops import UOp, Ops
+from tinygrad.ops import BinaryOps, MetaOps, UOp, UOps
 from tinygrad.shape.shapetracker import ShapeTracker

 # allocate some buffers + load in values
@@ -49,14 +49,14 @@ b = Buffer(DEVICE, 1, dtypes.int32).allocate().copyin(memoryview(bytearray(struc
 # NOTE: a._buf is the same as the return from MallocAllocator.alloc

 # describe the computation
-buf_1 = UOp(Ops.DEFINE_GLOBAL, dtypes.int32.ptr(), (), 1)
-buf_2 = UOp(Ops.DEFINE_GLOBAL, dtypes.int32.ptr(), (), 2)
-ld_1 = UOp(Ops.LOAD, dtypes.int32, (buf_1, ShapeTracker.from_shape((1,)).to_uop()))
-ld_2 = UOp(Ops.LOAD, dtypes.int32, (buf_2, ShapeTracker.from_shape((1,)).to_uop()))
+buf_1 = UOp(UOps.DEFINE_GLOBAL, PtrDType(dtypes.int32), (), 1)
+buf_2 = UOp(UOps.DEFINE_GLOBAL, PtrDType(dtypes.int32), (), 2)
+ld_1 = UOp(UOps.LOAD, dtypes.int32, (buf_1, ShapeTracker.from_shape((1,)).to_uop()))
+ld_2 = UOp(UOps.LOAD, dtypes.int32, (buf_2, ShapeTracker.from_shape((1,)).to_uop()))
 alu = ld_1 + ld_2
-output_buf = UOp(Ops.DEFINE_GLOBAL, dtypes.int32.ptr(), (), 0)
-st_0 = UOp(Ops.STORE, dtypes.void, (output_buf, ShapeTracker.from_shape((1,)).to_uop(), alu))
-s = UOp(Ops.SINK, dtypes.void, (st_0,))
+output_buf = UOp(UOps.DEFINE_GLOBAL, PtrDType(dtypes.int32), (), 0)
+st_0 = UOp(UOps.STORE, dtypes.void, (output_buf, ShapeTracker.from_shape((1,)).to_uop(), alu))
+s = UOp(UOps.SINK, dtypes.void, (st_0,))

 # convert the computation to a "linearized" format (print the format)
 from tinygrad.engine.realize import get_kernel, CompiledRunner
@@ -76,19 +76,20 @@ assert out.as_buffer().cast('I')[0] == 5

 print("******** third, the LazyBuffer ***********")

+from tinygrad.engine.lazy import LazyBuffer
 from tinygrad.engine.realize import run_schedule
 from tinygrad.engine.schedule import create_schedule

 # allocate some values + load in values
-a = UOp.metaop(Ops.EMPTY, (1,), dtypes.int32, DEVICE)
-b = UOp.metaop(Ops.EMPTY, (1,), dtypes.int32, DEVICE)
+a = LazyBuffer.metaop(MetaOps.EMPTY, (1,), dtypes.int32, DEVICE)
+b = LazyBuffer.metaop(MetaOps.EMPTY, (1,), dtypes.int32, DEVICE)
 a.buffer.allocate().copyin(memoryview(bytearray(struct.pack("I", 2))))
 b.buffer.allocate().copyin(memoryview(bytearray(struct.pack("I", 3))))
 del a.srcs
 del b.srcs

 # describe the computation
-out = a.alu(Ops.ADD, b)
+out = a.alu(BinaryOps.ADD, b)

 # schedule the computation as a list of kernels
 sched = create_schedule([out])
@@ -102,7 +103,7 @@ print(sched[-1].ast)
 run_schedule(sched)

 # check the data out
-assert out.realized is not None and out.realized.as_buffer().cast('I')[0] == 5
+assert out.realized.as_buffer().cast('I')[0] == 5


 print("******** fourth, the Tensor ***********")
@@ -1,6 +1,7 @@
 # abstractions2 goes from back to front, here we will go from front to back
 from typing import List
-from tinygrad.helpers import tqdm
+from tqdm import tqdm
+from tinygrad.helpers import DEBUG

 # *****
 # 0. Load mnist on the device
@@ -9,13 +9,17 @@ There is a good [bunch of tutorials](https://mesozoic-egg.github.io/tinygrad-not

 ## Frontend

-Everything in [Tensor](../tensor/index.md) is syntactic sugar around [function.py](function.md), where the forwards and backwards passes are implemented for the different functions. There's about 25 of them, implemented using about 20 basic ops. Those basic ops go on to construct a graph of [UOps](../developer/uop.md).
+Everything in [Tensor](../tensor/index.md) is syntactic sugar around [function.py](function.md), where the forwards and backwards passes are implemented for the different functions. There's about 25 of them, implemented using about 20 basic ops. Those basic ops go on to construct a graph of:

-The `UOp` graph specifies the compute in terms of low level tinygrad ops. Not all UOps will actually become realized. There's two types of UOps, base and view. base contains compute into a contiguous buffer, and view is a view (specified by a ShapeTracker). Inputs to a base can be either base or view, inputs to a view can only be a single base.
+::: tinygrad.engine.lazy.LazyBuffer
+    options:
+        show_source: false
+
+The `LazyBuffer` graph specifies the compute in terms of low level tinygrad ops. Not all LazyBuffers will actually become realized. There's two types of LazyBuffers, base and view. base contains compute into a contiguous buffer, and view is a view (specified by a ShapeTracker). Inputs to a base can be either base or view, inputs to a view can only be a single base.

 ## Scheduling

-The [scheduler](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/engine/schedule.py) converts the graph of UOps into a list of `ScheduleItem`. One `ScheduleItem` is one kernel on the GPU, and the scheduler is responsible for breaking the large compute graph into subgraphs that can fit in a kernel. `ast` specifies what compute to run, and `bufs` specifies what buffers to run it on.
+The [scheduler](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/engine/schedule.py) converts the graph of LazyBuffers into a list of `ScheduleItem`. One `ScheduleItem` is one kernel on the GPU, and the scheduler is responsible for breaking the large compute graph into subgraphs that can fit in a kernel. `ast` specifies what compute to run, and `bufs` specifies what buffers to run it on.

 ::: tinygrad.engine.schedule.ScheduleItem

@@ -6,29 +6,57 @@ The main aspect of HCQ-compatible runtimes is how they interact with devices. In

 ### Command Queues

-To interact with devices you create a `HWQueue`. Some methods are required, like timestamp and synchronization methods like [signal](#tinygrad.runtime.support.hcq.HWQueue.signal) and [wait](#tinygrad.runtime.support.hcq.HWQueue.wait), while others are dependent on it being a compute or copy queue.
+To interact with devices, there are 2 types of queues: `HWComputeQueue` and `HWCopyQueue`. Commands which are defined in a base `HWCommandQueue` class should be supported by both queues. These methods are timestamp and synchronization methods like [signal](#tinygrad.runtime.support.hcq.HWCommandQueue.signal) and [wait](#tinygrad.runtime.support.hcq.HWCommandQueue.wait).

 For example, the following Python code enqueues a wait, execute, and signal command on the HCQ-compatible device:
 ```python
-HWQueue().wait(signal_to_wait, value_to_wait) \
-         .exec(program, args_state, global_dims, local_dims) \
-         .signal(signal_to_fire, value_to_fire) \
-         .submit(your_device)
+HWComputeQueue().wait(signal_to_wait, value_to_wait) \
+                .exec(program, args_state, global_dims, local_dims) \
+                .signal(signal_to_fire, value_to_fire) \
+                .submit(your_device)
 ```

-Each runtime should implement the required functions that are defined in the `HWQueue` classes.
+Each runtime should implement the required functions that are defined in the `HWCommandQueue`, `HWComputeQueue`, and `HWCopyQueue` classes.

-::: tinygrad.runtime.support.hcq.HWQueue
+::: tinygrad.runtime.support.hcq.HWCommandQueue
    options:
        members: [
            "signal",
            "wait",
            "timestamp",
+            "update_signal",
+            "update_wait",
            "bind",
            "submit",
+        ]
+        show_source: false
+
+::: tinygrad.runtime.support.hcq.HWComputeQueue
+    options:
+        members: [
            "memory_barrier",
            "exec",
+            "update_exec",
+        ]
+        show_source: false
+
+::: tinygrad.runtime.support.hcq.HWCopyQueue
+    options:
+        members: [
            "copy",
+            "update_copy",
+        ]
+        show_source: false
+
+#### Implementing custom commands
+
+To implement custom commands in the queue, use the @hcq_command decorator for your command implementations.
+
+::: tinygrad.runtime.support.hcq.hcq_command
+    options:
+        members: [
+            "copy",
+            "update_copy",
        ]
        show_source: false

@@ -54,9 +82,9 @@ The following Python code demonstrates the usage of signals:
 ```python
 signal = your_device.signal_t()

-HWQueue().timestamp(signal) \
-         .signal(signal, value_to_fire) \
-         .submit(your_device)
+HWComputeQueue().timestamp(signal) \
+                .signal(signal, value_to_fire) \
+                .submit(your_device)

 signal.wait(value_to_fire)
 signaled_value = signal.value # should be the same as `value_to_fire`
@@ -106,17 +134,17 @@ Backends must adhere to the `HCQBuffer` protocol when returning allocation resul
        members: true
        show_source: false

-**Lifetime**: The `HCQArgsState` is passed to `HWQueue.exec` and is guaranteed not to be freed until `HWQueue.submit` for the same queue is called.
+**Lifetime**: The `HCQArgsState` is passed to `HWComputeQueue.exec` and is guaranteed not to be freed until `HWComputeQueue.submit` for the same queue is called.

 ### Synchronization

 HCQ-compatible devices use a global timeline signal for synchronizing all operations. This mechanism ensures proper ordering and completion of tasks across the device. By convention, `self.timeline_value` points to the next value to signal. So, to wait for all previous operations on the device to complete, wait for `self.timeline_value - 1` value. The following Python code demonstrates the typical usage of signals to synchronize execution to other operations on the device:

 ```python
-HWQueue().wait(your_device.timeline_signal, your_device.timeline_value - 1) \
-         .exec(...)
-         .signal(your_device.timeline_signal, your_device.timeline_value) \
-         .submit(your_device)
+HWComputeQueue().wait(your_device.timeline_signal, your_device.timeline_value - 1) \
+                .exec(...)
+                .signal(your_device.timeline_signal, your_device.timeline_value) \
+                .submit(your_device)
 your_device.timeline_value += 1

 # Optionally wait for execution
@@ -125,5 +153,5 @@ your_device.timeline_signal.wait(your_device.timeline_value - 1)

 ## HCQGraph

-[HCQGraph](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/graph/hcq.py) is a core feature that implements `GraphRunner` for HCQ-compatible devices. `HCQGraph` builds static `HWQueue` for all operations per device. To optimize enqueue time, only the necessary parts of the queues are updated for each run using the symbolic variables, avoiding a complete rebuild.
+[HCQGraph](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/runtime/graph/hcq.py) is a core feature that implements `GraphRunner` for HCQ-compatible devices. `HCQGraph` builds a static `HWComputeQueue` and `HWCopyQueue` for all operations per device. To optimize enqueue time, only the necessary parts of the queues are updated for each run using the update APIs of the queues, avoiding a complete rebuild.
 Optionally, queues can implement a `bind` API, which allows further optimization by eliminating the need to copy the queues into the device ring.
@@ -4,7 +4,7 @@
        members_order: source
        show_labels: false

-::: tinygrad.ops.Ops
+::: tinygrad.ops.UOps
    options:
        members: true
        members_order: source
@@ -40,11 +40,13 @@ METAL_XCODE         | [1]        | enable Metal using macOS Xcode SDK
 CLANG               | [1]        | enable Clang backend
 LLVM                | [1]        | enable LLVM backend
 BEAM                | [#]        | number of beams in kernel beam search
+GRAPH               | [1]        | create a graph of all operations (requires graphviz)
+GRAPHUOPS           | [1]        | create a graph of uops (requires graphviz and saves at /tmp/uops.{svg,dot})
+GRAPHPATH           | [/path/to] | where to put the generated graph
 DEFAULT_FLOAT       | [HALF, ...]| specify the default float dtype (FLOAT32, HALF, BFLOAT16, FLOAT64, ...), default to FLOAT32
 IMAGE               | [1-2]      | enable 2d specific optimizations
 FLOAT16             | [1]        | use float16 for images instead of float32
 PTX                 | [1]        | enable the specialized [PTX](https://docs.nvidia.com/cuda/parallel-thread-execution/) assembler for Nvidia GPUs. If not set, defaults to generic CUDA codegen backend.
 PROFILE             | [1]        | enable output of [perfetto](https://ui.perfetto.dev/) compatible profile. This feature is supported in NV and AMD backends.
 VISIBLE_DEVICES     | [list[int]]| restricts the NV/AMD devices that are available. The format is a comma-separated list of identifiers (indexing starts with 0).
-JIT                 | [0-2]      | 0=disabled, 1=[jit enabled](quickstart.md#jit) (default), 2=jit enabled, but graphs are disabled
-VIZ                 | [1]        | 0=disabled, 1=[viz enabled](https://github.com/tinygrad/tinygrad/tree/master/tinygrad/viz)
+JIT                 | [0-2]      | 0=disabled, 1=[jit enabled](quickstart.md#jit) (default), 2=jit enabled, but graphs are disabled
@@ -28,10 +28,6 @@ To understand what training looks like in tinygrad, you should read `beautiful_m

 We have a [quickstart guide](quickstart.md) and a [showcase](showcase.md)

-## tinygrad Stack
-
-<img src="./tinygrad_vs_others.png" alt="Tinygrad vs others" style="max-width: 1000px; height: auto;" />
-
 ## Differences from PyTorch

 If you are migrating from PyTorch, welcome. Most of the API is the same. We hope you will find tinygrad both familiar and somehow more "correct feeling"
@@ -165,7 +165,7 @@ from extra.datasets import fetch_mnist
 Now we have everything we need to start training our neural network.
 We will be training for 1000 steps with a batch size of 64.

-We use `with Tensor.train()` to set the internal flag `Tensor.training` to `True` during training.
+We use `with Tensor.train()` set the internal flag `Tensor.training` to `True` during training.
 Upon exit, the flag is restored to its previous value by the context manager.

 ```python
@@ -302,4 +302,7 @@ You can find a full list and their descriptions in [env_vars.md](env_vars.md).

 ### Visualizing the Computation Graph

-It is possible to visualize the computation graph of a neural network using VIZ=1.
+It is possible to visualize the computation graph of a neural network using [graphviz](https://graphviz.org/).
+
+This is easily done by running a single pass (forward or backward!) of the neural network with the environment variable `GRAPH` set to `1`.
+The graph will be saved to `/tmp/net.svg` by default.
@@ -5,16 +5,11 @@
 ::: tinygrad.Tensor.ones
 ::: tinygrad.Tensor.full
 ::: tinygrad.Tensor.arange
-::: tinygrad.Tensor.linspace
 ::: tinygrad.Tensor.eye
 ::: tinygrad.Tensor.full_like
 ::: tinygrad.Tensor.zeros_like
 ::: tinygrad.Tensor.ones_like
-
-## Creation (external)
-
 ::: tinygrad.Tensor.from_blob
-::: tinygrad.Tensor.from_url

 ## Creation (random)

@@ -13,15 +13,10 @@ Elementwise ops operate on a per element basis. They don't change the shape of t
 ::: tinygrad.Tensor.sin
 ::: tinygrad.Tensor.cos
 ::: tinygrad.Tensor.tan
-::: tinygrad.Tensor.asin
-::: tinygrad.Tensor.acos
-::: tinygrad.Tensor.atan
 ::: tinygrad.Tensor.trunc
 ::: tinygrad.Tensor.ceil
 ::: tinygrad.Tensor.floor
 ::: tinygrad.Tensor.round
-::: tinygrad.Tensor.isinf
-::: tinygrad.Tensor.isnan
 ::: tinygrad.Tensor.lerp
 ::: tinygrad.Tensor.square
 ::: tinygrad.Tensor.clamp
@@ -34,10 +29,8 @@ Elementwise ops operate on a per element basis. They don't change the shape of t

 ::: tinygrad.Tensor.relu
 ::: tinygrad.Tensor.sigmoid
-::: tinygrad.Tensor.hardsigmoid
 ::: tinygrad.Tensor.elu
 ::: tinygrad.Tensor.celu
-::: tinygrad.Tensor.selu
 ::: tinygrad.Tensor.swish
 ::: tinygrad.Tensor.silu
 ::: tinygrad.Tensor.relu6
@@ -49,7 +42,6 @@ Elementwise ops operate on a per element basis. They don't change the shape of t
 ::: tinygrad.Tensor.asinh
 ::: tinygrad.Tensor.acosh
 ::: tinygrad.Tensor.hardtanh
-::: tinygrad.Tensor.erf
 ::: tinygrad.Tensor.gelu
 ::: tinygrad.Tensor.quick_gelu
 ::: tinygrad.Tensor.leakyrelu
@@ -17,9 +17,9 @@
 ::: tinygrad.Tensor.repeat_interleave
 ::: tinygrad.Tensor.split
 ::: tinygrad.Tensor.chunk
-::: tinygrad.Tensor.meshgrid
 ::: tinygrad.Tensor.squeeze
 ::: tinygrad.Tensor.unsqueeze
+::: tinygrad.Tensor.pad2d
 ::: tinygrad.Tensor.T
 ::: tinygrad.Tensor.transpose
 ::: tinygrad.Tensor.flatten
@@ -27,11 +27,9 @@
 ::: tinygrad.Tensor.matmul
 ::: tinygrad.Tensor.einsum
 ::: tinygrad.Tensor.cumsum
-::: tinygrad.Tensor.cummax
 ::: tinygrad.Tensor.triu
 ::: tinygrad.Tensor.tril
 ::: tinygrad.Tensor.interpolate
-::: tinygrad.Tensor.scatter

 ## Neural Network (functional)

@@ -46,4 +44,3 @@
 ::: tinygrad.Tensor.binary_crossentropy_logits
 ::: tinygrad.Tensor.sparse_categorical_crossentropy
 ::: tinygrad.Tensor.cross_entropy
-::: tinygrad.Tensor.nll_loss
@@ -1,9 +0,0 @@
-import globals from "globals";
-import pluginJs from "@eslint/js";
-import pluginHtml from "eslint-plugin-html";
-
-export default [
-  {files: ["**/*.html"], plugins: {html: pluginHtml}, rules:{"max-len": ["error", {"code": 150}]}},
-  {languageOptions: {globals: globals.browser}},
-  pluginJs.configs.recommended,
-];
@@ -1,7 +1,7 @@
 from pathlib import Path
 from extra.models.efficientnet import EfficientNet
 from tinygrad.tensor import Tensor
-from tinygrad.nn.state import get_state_dict, safe_save, safe_load, load_state_dict
+from tinygrad.nn.state import safe_save
 from extra.export_model import export_model
 from tinygrad.helpers import getenv, fetch
 import ast
@@ -9,16 +9,12 @@ import ast
 if __name__ == "__main__":
  model = EfficientNet(0)
  model.load_from_pretrained()
-  dirname = Path(__file__).parent
-  # exporting a model that's loaded from safetensors doesn't work without loading in from safetensors first
-  # loading the state dict from a safetensor file changes the generated kernels
-  if getenv("WEBGPU"):
-    safe_save(get_state_dict(model), (dirname / "net.safetensors").as_posix())
-    load_state_dict(model, safe_load(str(dirname / "net.safetensors")))
-  mode = "clang" if getenv("CLANG", "") != "" else "webgpu" if getenv("WEBGPU", "") != "" else ""
+  mode = "clang" if getenv("CLANG", "") != "" else "webgpu" if getenv("WEBGPU", "") != "" else "webgl" if getenv("WEBGL", "") != "" else ""
  prg, inp_sizes, out_sizes, state = export_model(model, mode, Tensor.randn(1,3,224,224))
+  dirname = Path(__file__).parent
  if getenv("CLANG", "") == "":
-    ext = "js" if getenv("WEBGPU", "") != "" else "json"
+    safe_save(state, (dirname / "net.safetensors").as_posix())
+    ext = "js" if getenv("WEBGPU", "") != "" or getenv("WEBGL", "") != "" else "json"
    with open(dirname / f"net.{ext}", "w") as text_file:
      text_file.write(prg)
  else:
@@ -326,12 +326,11 @@ class Flux:
    return self.final_layer(img, vec)  # (N, T, patch_size ** 2 * out_channels)

 # https://github.com/black-forest-labs/flux/blob/main/src/flux/util.py
-def load_flow_model(name:str, model_path:str):
+def load_flow_model(name:str):
  # Loading Flux
  print("Init model")
  model = Flux(guidance_embed=(name != "flux-schnell"))
-  if not model_path: model_path = fetch(urls[name])
-  state_dict = {k.replace("scale", "weight"): v for k, v in safe_load(model_path).items()}
+  state_dict = {k.replace("scale", "weight"): v for k, v in safe_load(fetch(urls[name])).items()}
  load_state_dict(model, state_dict)
  return model

@@ -421,7 +420,6 @@ if __name__ == "__main__":
  parser = argparse.ArgumentParser(description="Run Flux.1", formatter_class=argparse.ArgumentDefaultsHelpFormatter)

  parser.add_argument("--name",       type=str,   default="flux-schnell", help="Name of the model to load")
-  parser.add_argument("--model_path", type=str,   default="",             help="path of the model file")
  parser.add_argument("--width",      type=int,   default=512,            help="width of the sample in pixels (should be a multiple of 16)")
  parser.add_argument("--height",     type=int,   default=512,            help="height of the sample in pixels (should be a multiple of 16)")
  parser.add_argument("--seed",       type=int,   default=None,           help="Set a seed for sampling")
@@ -463,7 +461,7 @@ if __name__ == "__main__":
  del T5, clip

  # load model
-  model = load_flow_model(args.name, args.model_path)
+  model = load_flow_model(args.name)

  # denoise initial noise
  x = denoise(model, **inp, timesteps=timesteps, guidance=args.guidance)
@@ -1,12 +1,13 @@
 #!/usr/bin/env python3
-import os, argparse
 from typing import Optional, Union
+import argparse
+import numpy as np
 import tiktoken
-from tinygrad import Tensor, TinyJit, Device, GlobalCounters, Variable, dtypes
+from tinygrad import Tensor, TinyJit, Device, GlobalCounters, Variable
 from tinygrad.ops import UOp
 from tinygrad.helpers import Timing, DEBUG, JIT, getenv, fetch, colored, trange
 from tinygrad.nn import Embedding, Linear, LayerNorm
-from tinygrad.nn.state import gguf_load, torch_load, load_state_dict, get_state_dict
+from tinygrad.nn.state import torch_load, load_state_dict, get_state_dict

 MAX_CONTEXT = getenv("MAX_CONTEXT", 128)
 HALF = getenv("HALF")
@@ -142,34 +143,6 @@ class GPT2:

    return GPT2(model, tokenizer)

-  @staticmethod
-  def build_gguf(model_size: str):
-    q_type = model_size[len("gpt2_gguf_"):].upper()
-    fn = fetch(f"https://huggingface.co/PrunaAI/gpt2-GGUF-smashed/resolve/main/gpt2.{q_type}.gguf?download=true")
-    gguf_tensor = Tensor.empty(os.stat(fn).st_size, dtype=dtypes.uint8, device=f"disk:{fn}").to(Device.DEFAULT)
-    kv_data, state_dict = gguf_load(gguf_tensor)
-
-    gpt2_params = {
-      "dim": kv_data["gpt2.embedding_length"], "n_heads": kv_data["gpt2.attention.head_count"],
-      "n_layers": kv_data["gpt2.block_count"], "norm_eps": kv_data["gpt2.attention.layer_norm_epsilon"],
-      "vocab_size": VOCAB_SIZE, "max_seq_len": kv_data["gpt2.context_length"],
-    }
-    def _remap_gguf_key(key: str):
-      replaces = [
-        ("blk.", "h."), (".attn_qkv.bias", ".attn.c_attn.bias"), (".attn_qkv.weight", ".attn.c_attn.weight"),
-        (".ffn_norm.bias", ".ln_2.bias"), (".ffn_norm.weight", ".ln_2.weight"), (".attn_norm.bias", ".ln_1.bias"),
-        (".attn_norm.weight", ".ln_1.weight"), (".attn_output.bias", ".attn.c_proj.bias"), (".attn_output.weight", ".attn.c_proj.weight"),
-        (".ffn_up.bias", ".mlp.c_fc.bias"), (".ffn_up.weight", ".mlp.c_fc.weight"), (".ffn_down.bias", ".mlp.c_proj.bias"),
-        (".ffn_down.weight", ".mlp.c_proj.weight"), ("token_embd.weight", "wte.weight"), ("output.weight", "lm_head.weight"),
-        ("output_norm.bias", "ln_f.bias"), ("output_norm.weight", "ln_f.weight"), ("position_embd.weight", "wpe.weight"),
-      ]
-      for ostr, ns in replaces: key = key.replace(ostr, ns)
-      return key
-    state_dict = { _remap_gguf_key(k): v for k, v in state_dict.items() }
-    model = Transformer(**gpt2_params)
-    load_state_dict(model, state_dict)
-    return GPT2(model, tiktoken.get_encoding("gpt2"))
-
  def __init__(self, model, tokenizer):
    self.model = model
    self.tokenizer = tokenizer
@@ -189,7 +162,7 @@ class GPT2:
          tokens = Variable("tokens", 0, VOCAB_SIZE).bind(toks[0][start_pos])
        else:
          tokens = Tensor([x[start_pos:] for x in toks])
-        tok = self.model(tokens, Variable("start_pos", 1 if start_pos else 0, MAX_CONTEXT).bind(start_pos), temperature).tolist()
+        tok = self.model(tokens, Variable("start_pos", 1 if start_pos else 0, MAX_CONTEXT).bind(start_pos), temperature).numpy().tolist()
      start_pos = len(toks[0])
      for i,t in enumerate(tok): toks[i].append(t)
    return [self.tokenizer.decode(x) for x in toks]
@@ -215,9 +188,10 @@ if __name__ == "__main__":

  if args.seed is not None:
    Tensor.manual_seed(args.seed)
+    np.random.seed(args.seed)

  print(f"using {args.model_size}")
-  gpt2 = GPT2.build_gguf(args.model_size) if args.model_size.startswith("gpt2_gguf_") else GPT2.build(args.model_size)
+  gpt2 = GPT2.build(args.model_size)

  if args.benchmark != -1:
    gpt2.model(Tensor.rand(args.batch_size, args.benchmark), Variable("a", 0, MAX_CONTEXT).bind(0)).realize()
@@ -4,11 +4,12 @@ from extra.mcts_search import mcts_search
 from examples.mlperf.helpers import get_mlperf_bert_model
 from tinygrad import Tensor, Device, dtypes, nn
 from tinygrad.codegen.kernel import Kernel
-from tinygrad.ops import Ops, sym_infer
+from tinygrad.ops import UOps
 from tinygrad.device import Compiled
 from tinygrad.engine.schedule import create_schedule
 from tinygrad.engine.search import time_linearizer, beam_search, bufs_from_lin
 from tinygrad.helpers import DEBUG, ansilen, getenv, colored, TRACEMETA
+from tinygrad.shape.symbolic import sym_infer

 def get_sched_resnet():
  mdl = ResNet50()
@@ -66,7 +67,7 @@ if __name__ == "__main__":
  print(f"optimizing for {Device.DEFAULT}")

  sched = globals()[f"get_sched_{getenv('MODEL', 'resnet')}"]()
-  sched = [x for x in sched if x.ast.op is Ops.SINK]
+  sched = [x for x in sched if x.ast.op is UOps.SINK]

  # focus on one kernel
  if getenv("KERNEL", -1) >= 0: sched = sched[getenv("KERNEL", -1):getenv("KERNEL", -1)+1]
@@ -125,9 +126,9 @@ if __name__ == "__main__":
      print(lin.to_program().src)
    total_tm += tm
    running_gflops += gflops * tm
-    if (key := str([str(m) for m in si.metadata])) not in usage: usage[key] = (0, 0)
+    if (key := str([str(m) for m in si.metadata] if si.metadata is not None else None)) not in usage: usage[key] = (0, 0)
    usage[key] = (usage[key][0] + tm, usage[key][1] + 1)
-    print(f"*** {total_tm*1000:7.2f} ms : kernel {i:2d} {lin.name+' '*(37-ansilen(lin.name))} {str(prg.global_size):18s} {str(prg.local_size):12s} takes {tm*1000:7.2f} ms, {gflops:6.0f} GFLOPS {[repr(m) if TRACEMETA >= 2 else str(m) for m in si.metadata]}")
+    print(f"*** {total_tm*1000:7.2f} ms : kernel {i:2d} {lin.name+' '*(37-ansilen(lin.name))} {str(prg.global_size):18s} {str(prg.local_size):12s} takes {tm*1000:7.2f} ms, {gflops:6.0f} GFLOPS {[repr(m) if TRACEMETA >= 2 else str(m) for m in si.metadata] if si.metadata is not None else ''}")
  print(f"******* total {total_tm*1000:.2f} ms, {running_gflops/total_tm:6.0f} GFLOPS")
  print("usage:")
  for k in sorted(usage, key=lambda x: -usage[x][0])[:10]:
@@ -114,7 +114,7 @@ class SpeedyResNet:
  def __call__(self, x, training=True):
    # pad to 32x32 because whitening conv creates 31x31 images that are awfully slow to compute with
    # TODO: remove the pad but instead let the kernel optimize itself
-    forward = lambda x: x.conv2d(self.whitening).pad((1,0,0,1)).sequential(self.net)
+    forward = lambda x: x.conv2d(self.whitening).pad2d((1,0,0,1)).sequential(self.net)
    return forward(x) if training else (forward(x) + forward(x[..., ::-1])) / 2.

 # hyper-parameters were exactly the same as the original repo
@@ -17,11 +17,8 @@ canvas {  display: none; }
 * {  text-align: center;  font-family: monospace; }
 </style>
 <title>tinygrad has WebGPU</title>
+<script src="./net.js"></script>
 <link rel="icon" type="image/x-icon" href="https://raw.githubusercontent.com/tinygrad/tinygrad/master/docs/logo.png">
-<script type="module">
-	import model from "../../net.js";
-	window.model = model;
-</script>
 </head>
 <body>
 <h1>WebGPU <a href="https://github.com/geohot/tinygrad">tinygrad</a> EfficientNet!</h1>
@@ -64,6 +61,8 @@ canvas {  display: none; }

 	const getLabels = async () => (await fetch("https://raw.githubusercontent.com/anishathalye/imagenet-simple-labels/master/imagenet-simple-labels.json")).json();

+	const getSavetensorBuffer = async () => new Uint8Array(await (await fetch("./net.safetensors")).arrayBuffer());
+
 	const reorderChannelsAndRemoveAlpha = (data) => {
 		const out = [];
 		let i = 0;
@@ -98,8 +97,9 @@ canvas {  display: none; }
 		try {
 			resultText.innerHTML = "loading..."
 			labels = await getLabels();
+			const safetensor = await getSavetensorBuffer();
 			const device = await getDevice();
-			net = await timer(() => model.load(device, '../../net.safetensors'), "(compilation)");
+			net = await timer(() => setupNet(device, safetensor), "(compilation)");
 			resultText.innerHTML = "ready"
 		} catch (e) {
 			error(e)
@@ -6,6 +6,8 @@
 from pathlib import Path
 from typing import List, Optional
 import argparse, json
+import numpy as np
+np.set_printoptions(linewidth=200)
 from tinygrad import Tensor, Device, GlobalCounters, nn
 from tinygrad.helpers import Context, Timing, Profiling, DEBUG, JIT, getenv, colored
 from tinygrad.nn.state import safe_load, torch_load, load_state_dict, get_parameters
@@ -248,8 +250,6 @@ class LLaMa:
    self.tokenizer = tokenizer

  def greedy_until(self, prompt:str, until, max_length, temperature):
-    # only used in old eval script
-    import numpy as np
    toks = [self.tokenizer.bos_id()] + self.tokenizer.encode(prompt)
    start_pos = 0
    for i in range(max_length):
@@ -498,16 +498,14 @@ After you are done speaking, output [EOS]. You are not Chad.
    if not chatbot: break

  # validate output!
-  if args.temperature == 0 and args.count == 10 and args.prompt == "Hello.":
+  if args.temperature == 0 and args.count == 10 and args.prompt == "Hello." and not args.quantize:
    text = llama.tokenizer.decode(toks)
-    key = (args.gen, args.size, args.quantize)
+    key = (args.gen, args.size)
    expected = {
-      ("1", "7B", None): "Hello. I'm a 20 year old male",
-      ("1", "7B", "int8"): "Hello. I'm a 20 year old male",
-      ("1", "7B", "nf4"): "Hello. I'm a 20 year old male",
-      ("2", "7B", None): "Hello. I'm a 20 year old girl",
-      ("2", "70B", None): "Hello. I am a 20 year old female.",
-      ("3", "8B", None): "Hello. I am a 20 year old female. I",
+      ("1", "7B"): "Hello. I'm a 20 year old male",
+      ("2", "7B"): "Hello. I'm a 20 year old girl",
+      ("2", "70B"): "Hello. I am a 20 year old female.",
+      ("3", "8B"): "Hello. I am a 20 year old female. I",
    }
    try:
      assert text == expected[key], f"invalid output: `{colored(text, 'red')}` != `{expected[key]}`"
@@ -1,10 +1,10 @@
 from pathlib import Path
 from typing import List
-import json, argparse, random, time, os
+import json, argparse, random, time
 import tiktoken
 from tiktoken.load import load_tiktoken_bpe
-from extra.models.llama import Transformer, convert_from_huggingface, convert_from_gguf, fix_bf16
-from tinygrad.nn.state import safe_load, torch_load, load_state_dict, get_parameters, gguf_load
+from extra.models.llama import Transformer, convert_from_huggingface, fix_bf16
+from tinygrad.nn.state import safe_load, torch_load, load_state_dict, get_parameters
 from tinygrad import Tensor, dtypes, nn, Context, Device, GlobalCounters
 from tinygrad.helpers import Profiling, Timing, DEBUG, colored, fetch, tqdm

@@ -57,9 +57,6 @@ def load(fn:str):
    with open(fn) as fp: weight_map = json.load(fp)['weight_map']
    parts = {n: load(str(Path(fn).parent / Path(n).name)) for n in set(weight_map.values())}
    return {k: parts[n][k] for k, n in weight_map.items()}
-  elif fn.endswith(".gguf"):
-    gguf_tensor = Tensor.empty(os.stat(fn).st_size, dtype=dtypes.uint8, device=f"disk:{fn}").to(Device.DEFAULT)
-    return gguf_load(gguf_tensor)[1]
  elif fn.endswith(".safetensors"):
    return safe_load(fn)
  else:
@@ -108,7 +105,7 @@ def NF4Linear(block_size):
    def __call__(self, x: Tensor) -> Tensor:
      high_bits = self.weight
      low_bits = (self.weight * 2 ** 4).contiguous()
-      unpacked = Tensor.stack(high_bits, low_bits, dim=-1).idiv(2 ** 4)
+      unpacked = Tensor.stack(high_bits, low_bits, dim=-1).div(2 ** 4, upcast=False)
      unscaled = CODE[unpacked].to(x.device).reshape(-1, block_size) * self.scale
      return x.linear(unscaled.reshape(self.out_features, self.in_features).T)

@@ -131,10 +128,6 @@ def NF4Linear(block_size):
  return _NF4Linear

 MODEL_PARAMS = {
-  "1B": {
-    "args": {"dim": 2048, "n_heads": 32, "n_kv_heads": 8, "n_layers": 16, "norm_eps": 1e-5, "rope_theta": 500000, "vocab_size": 128256, "hidden_dim": 8192},
-    "files": 1
-  },
  "8B": {
    "args": {"dim": 4096, "n_heads": 32, "n_kv_heads": 8, "n_layers": 32, "norm_eps": 1e-5, "rope_theta": 500000, "vocab_size": 128256, "hidden_dim": 14336},
    "files": 1
@@ -160,14 +153,11 @@ def build_transformer(model_path: Path, model_size="8B", quantize=None, device=N
    weights = load(str(model_path))
  if "model.embed_tokens.weight" in weights:
    weights = convert_from_huggingface(weights, model, MODEL_PARAMS[model_size]["args"]["n_heads"], MODEL_PARAMS[model_size]["args"]["n_kv_heads"])
-  elif "token_embd.weight" in weights:
-    weights = convert_from_gguf(weights, model)
  weights = fix_bf16(weights)

  with Context(BEAM=0):
    # quantize
-    if quantize == "float16": weights = {k:v.cast(quantize).contiguous() for k,v in weights.items()}
-    elif quantize is not None:
+    if quantize is not None:
      weights = linear.quantize(weights, device)
      for _,v in weights.items(): v.realize()

@@ -218,11 +208,11 @@ if __name__ == "__main__":
  Tensor.no_grad = True

  parser = argparse.ArgumentParser()
-  parser.add_argument("--download_model", action="store_true", help="Download a model")
+  parser.add_argument("--download_model", action="store_true", help="Download a 8B model")
  parser.add_argument("--model", type=Path, help="Model path")
-  parser.add_argument("--size", choices=["1B", "8B", "70B"], default="1B", help="Model size")
+  parser.add_argument("--size", choices=["8B", "70B"], default="8B", help="Model size")
  parser.add_argument("--shard", type=int, default=1, help="Shard the model across multiple devices")
-  parser.add_argument("--quantize", choices=["int8", "nf4", "float16"], help="Quantization method")
+  parser.add_argument("--quantize", choices=["int8", "nf4"], help="Quantization method")
  parser.add_argument("--no_api", action="store_true", help="Disable the api and run a cli test interface")
  parser.add_argument("--host", type=str, default="0.0.0.0", help="Web server bind address")
  parser.add_argument("--port", type=int, default=7776, help="Web server port")
@@ -234,24 +224,14 @@ if __name__ == "__main__":
  parser.add_argument("--profile", action="store_true", help="Output profile data")
  args = parser.parse_args()

-  # download_model is the default without a model passed in
-  if args.download_model or not args.model:
-    if args.size == "1B":
-      fetch("https://huggingface.co/bofenghuang/Meta-Llama-3-8B/resolve/main/original/tokenizer.model", "tokenizer.model", subdir="llama3-1b-instruct")
-      args.model = fetch("https://huggingface.co/bartowski/Llama-3.2-1B-Instruct-GGUF/resolve/main/Llama-3.2-1B-Instruct-Q6_K.gguf", "Llama-3.2-1B-Instruct-Q6_K.gguf", subdir="llama3-1b-instruct")
-    elif args.size == "8B":
-      fetch("https://huggingface.co/bofenghuang/Meta-Llama-3-8B/resolve/main/original/tokenizer.model", "tokenizer.model", subdir="llama3-8b-sfr")
-      fetch("https://huggingface.co/TriAiExperiments/SFR-Iterative-DPO-LLaMA-3-8B-R/resolve/main/model-00001-of-00004.safetensors", "model-00001-of-00004.safetensors", subdir="llama3-8b-sfr")
-      fetch("https://huggingface.co/TriAiExperiments/SFR-Iterative-DPO-LLaMA-3-8B-R/resolve/main/model-00002-of-00004.safetensors", "model-00002-of-00004.safetensors", subdir="llama3-8b-sfr")
-      fetch("https://huggingface.co/TriAiExperiments/SFR-Iterative-DPO-LLaMA-3-8B-R/resolve/main/model-00003-of-00004.safetensors", "model-00003-of-00004.safetensors", subdir="llama3-8b-sfr")
-      fetch("https://huggingface.co/TriAiExperiments/SFR-Iterative-DPO-LLaMA-3-8B-R/resolve/main/model-00004-of-00004.safetensors", "model-00004-of-00004.safetensors", subdir="llama3-8b-sfr")
-      args.model = fetch("https://huggingface.co/TriAiExperiments/SFR-Iterative-DPO-LLaMA-3-8B-R/raw/main/model.safetensors.index.json", "model.safetensors.index.json", subdir="llama3-8b-sfr")
-    elif args.size == "70B":
-      subdir = "Llama-3.1-Nemotron-70B-Instruct-HF"
-      args.model = fetch("https://huggingface.co/nvidia/Llama-3.1-Nemotron-70B-Instruct-HF/resolve/main/model.safetensors.index.json?download=true", "model.safetensors.index.json", subdir=subdir)
-      fetch("https://huggingface.co/bofenghuang/Meta-Llama-3-8B/resolve/main/original/tokenizer.model", "tokenizer.model", subdir=subdir)
-      for i in range(30):
-        fetch(f"https://huggingface.co/nvidia/Llama-3.1-Nemotron-70B-Instruct-HF/resolve/main/model-{i+1:05d}-of-00030.safetensors?download=true", f"model-{i+1:05d}-of-00030.safetensors", subdir=subdir)
+  assert not (args.download_model and args.model), "either download or provide model"
+  if args.download_model:
+    fetch("https://huggingface.co/bofenghuang/Meta-Llama-3-8B/resolve/main/original/tokenizer.model", "tokenizer.model", subdir="llama3-8b-sfr")
+    fetch("https://huggingface.co/TriAiExperiments/SFR-Iterative-DPO-LLaMA-3-8B-R/resolve/main/model-00001-of-00004.safetensors", "model-00001-of-00004.safetensors", subdir="llama3-8b-sfr")
+    fetch("https://huggingface.co/TriAiExperiments/SFR-Iterative-DPO-LLaMA-3-8B-R/resolve/main/model-00002-of-00004.safetensors", "model-00002-of-00004.safetensors", subdir="llama3-8b-sfr")
+    fetch("https://huggingface.co/TriAiExperiments/SFR-Iterative-DPO-LLaMA-3-8B-R/resolve/main/model-00003-of-00004.safetensors", "model-00003-of-00004.safetensors", subdir="llama3-8b-sfr")
+    fetch("https://huggingface.co/TriAiExperiments/SFR-Iterative-DPO-LLaMA-3-8B-R/resolve/main/model-00004-of-00004.safetensors", "model-00004-of-00004.safetensors", subdir="llama3-8b-sfr")
+    args.model = fetch("https://huggingface.co/TriAiExperiments/SFR-Iterative-DPO-LLaMA-3-8B-R/raw/main/model.safetensors.index.json", "model.safetensors.index.json", subdir="llama3-8b-sfr")

  assert args.model is not None, "please provide --model option"

@@ -288,9 +268,8 @@ if __name__ == "__main__":
      for key, value in cors_headers.items(): response.set_header(key, value)

    @app.route("/<filename>")
-    def server_static(filename): return static_file(filename, root=(Path(__file__).parent / "tinychat").as_posix())
-    @app.route("/assets/<filename:path>")
-    def server_assets(filename): return static_file(filename, root=(Path(__file__).parent / "tinychat" / "assets").as_posix())
+    def server_static(filename):
+      return static_file(filename, root=(Path(__file__).parent / "tinychat").as_posix())
    @app.route("/")
    def index():
      return static_file("index.html", root=(Path(__file__).parent / "tinychat").as_posix())
@@ -4,11 +4,10 @@ if "NOOPT" not in os.environ: os.environ["NOOPT"] = "1"
 from tinygrad import Device, nn, Tensor, dtypes, Variable
 Device.DEFAULT = "CLANG"
 from train_gpt2 import GPT, GPTConfig
-from tinygrad.helpers import dedup, to_function_name, flatten, getenv, GlobalCounters, ansilen, to_function_name
+from tinygrad.helpers import dedup, to_function_name, flatten, getenv, GRAPH, GlobalCounters, ansilen, to_function_name
 from tinygrad.engine.schedule import create_schedule
-from tinygrad.engine.realize import get_kernel, run_schedule
-from tinygrad.engine.memory import memory_planner
-from tinygrad.ops import Ops
+from tinygrad.engine.realize import get_kernel, memory_planner, run_schedule
+from tinygrad.ops import MetaOps, UOps

 TIMING = getenv("TIMING")

@@ -27,6 +26,7 @@ if __name__ == "__main__":
  optimizer = nn.optim.Adam(nn.state.get_parameters(model), lr=1e-4)
  warmup_count = getenv("WARMUP", 3)
  for i in range(warmup_count):  # TODO: why does it take three and not two to stablize
+    if i == warmup_count-1: GRAPH.value = getenv("LATEGRAPH")
    GlobalCounters.reset()
    X = Tensor.empty(4, 64, dtype=dtypes.int).reshape(B, T)
    Y = Tensor.empty(4, 64, dtype=dtypes.int).reshape(B, T)
@@ -41,7 +41,7 @@ if __name__ == "__main__":
    print(f"calls {i}:", len(sched))
    #run_schedule(sched[:])
  sched = memory_planner(sched)
-  ast_dedup = dedup([si.ast for si in sched if si.ast.op is Ops.SINK])
+  ast_dedup = dedup([si.ast for si in sched if si.ast.op is UOps.SINK])
  srcs = {}
  for ast in ast_dedup:
    k = get_kernel(Device["CLANG"].renderer, ast)
@@ -82,7 +82,7 @@ if __name__ == "__main__":
  for i,si in enumerate(sched):
    bufs = [(named_buffers.get(b, f"b{numbered_bufs[b]}"), b) for b in si.bufs]
    all_bufs += bufs
-    if si.ast.op is not Ops.SINK:
+    if si.ast.op is not UOps.SINK:
      print(f"// {si.ast.op}", bufs)
    else:
      print(f"{srcs[si.ast][0]}({', '.join([x[0] for x in bufs])})")
@@ -67,11 +67,11 @@ def loader_process(q_in, q_out, X:Tensor, seed):

      # broken out
      #img_tensor = Tensor(img.tobytes(), device='CPU')
-      #storage_tensor = X[idx].contiguous().realize().lazydata.base.realized
+      #storage_tensor = X[idx].contiguous().realize().lazydata.realized
      #storage_tensor._copyin(img_tensor.numpy())

      # faster
-      X[idx].contiguous().realize().lazydata.base.realized.as_buffer(force_zero_copy=True)[:] = img.tobytes()
+      X[idx].contiguous().realize().lazydata.realized.as_buffer(force_zero_copy=True)[:] = img.tobytes()

      # ideal
      #X[idx].assign(img.tobytes())   # NOTE: this is slow!
@@ -267,8 +267,8 @@ def load_unet3d_data(preprocessed_dataset_dir, seed, queue_in, queue_out, X:Tens
      x = random_brightness_augmentation(x)
      x = gaussian_noise(x)

-    X[idx].contiguous().realize().lazydata.base.realized.as_buffer(force_zero_copy=True)[:] = x.tobytes()
-    Y[idx].contiguous().realize().lazydata.base.realized.as_buffer(force_zero_copy=True)[:] = y.tobytes()
+    X[idx].contiguous().realize().lazydata.realized.as_buffer(force_zero_copy=True)[:] = x.tobytes()
+    Y[idx].contiguous().realize().lazydata.realized.as_buffer(force_zero_copy=True)[:] = y.tobytes()

    queue_out.put(idx)
  queue_out.put(None)
@@ -1,5 +1,6 @@
 import os, time, math, functools
 from pathlib import Path
+from tqdm import tqdm
 import multiprocessing

 from tinygrad import Device, GlobalCounters, Tensor, TinyJit, dtypes
@@ -9,8 +10,6 @@ from tinygrad.nn.optim import LAMB, LARS, SGD, OptimizerGroup

 from extra.lr_scheduler import LRSchedulerGroup
 from examples.mlperf.helpers import get_training_state, load_training_state
-# TODO: fix benchmark logging and use tinygrad tqdm
-from tqdm import tqdm

 def train_resnet():
  from extra.models import resnet
@@ -642,7 +641,6 @@ def train_bert():

    if RUNMLPERF:
      MLLOGGER.start(key=mllog_constants.RUN_START, value=None)
-      MLLOGGER.event(key=mllog_constants.SEED, value=seed)
  else:
    MLLOGGER = None

@@ -1,50 +0,0 @@
-# 1. Problem
-
-This problem uses the ResNet-50 CNN to do image classification.
-
-## Requirements
-
-Install tinygrad and mlperf-logging from master.
-```
-git clone https://github.com/tinygrad/tinygrad.git
-python3 -m pip install -e ".[mlperf]"
-```
-
-### tinybox_green
-Install the p2p driver per [README](https://github.com/tinygrad/open-gpu-kernel-modules/blob/550.54.15-p2p/README.md)
-This is the default on production tinybox green.
-
-### tinybox_red
-Disable cwsr
-This is the default on production tinybox red.
-```
-sudo vi /etc/modprobe.d/amdgpu.conf
-cat <<EOF > /etc/modprobe.d/amdgpu.conf
-options amdgpu cwsr_enable=0
-EOF
-sudo update-initramfs -u
-sudo reboot
-
-# validate
-sudo cat /sys/module/amdgpu/parameters/cwsr_enable #= 0
-```
-
-# 2. Directions
-
-## Steps to download and verify data
-
-```
-IMGNET_TRAIN=1 python3 extra/datasets/imagenet_download.py
-```
-
-## Steps for one time setup
-
-### tinybox_red
-```
-examples/mlperf/training_submission_v4.0/tinycorp/benchmarks/resnet/implementations/tinybox_red/setup.sh
-```
-
-## Steps to run benchmark
-```
-examples/mlperf/training_submission_v4.0/tinycorp/benchmarks/resnet/implementations/tinybox_red/run_and_time.sh
-```
@@ -1,13 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="resnet"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=1536 EVAL_BS=192
-
-export LAZYCACHE=0 RESET_STEP=0
-
-export TRAIN_BEAM=4 IGNORE_JIT_FIRST_BEAM=1 BEAM_UOPS_MAX=1500 BEAM_UPCAST_MAX=64 BEAM_LOCAL_MAX=1024 BEAM_MIN_PROGRESS=10 BEAM_PADTO=0
-
-export BENCHMARK=10 DEBUG=2
-
-python3 examples/mlperf/model_train.py
@@ -1,15 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="resnet"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=1536 EVAL_BS=192
-
-export LAZYCACHE=0 RESET_STEP=0
-
-export TRAIN_BEAM=4 IGNORE_JIT_FIRST_BEAM=1 BEAM_UOPS_MAX=1500 BEAM_UPCAST_MAX=64 BEAM_LOCAL_MAX=1024 BEAM_MIN_PROGRESS=10 BEAM_PADTO=0
-
-export EVAL_START_EPOCH=3 EVAL_FREQ=4
-
-export WANDB=1 PARALLEL=0
-
-python3 examples/mlperf/model_train.py
@@ -1,23 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="resnet"
-export SUBMISSION_PLATFORM="tinybox_green"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=1536 EVAL_BS=192
-
-export LAZYCACHE=0 RESET_STEP=0
-
-export TRAIN_BEAM=4 IGNORE_JIT_FIRST_BEAM=1 BEAM_UOPS_MAX=1500 BEAM_UPCAST_MAX=64 BEAM_LOCAL_MAX=1024 BEAM_MIN_PROGRESS=10 BEAM_PADTO=0
-
-# pip install -e ".[mlperf]"
-export LOGMLPERF=1
-
-export SEED=$RANDOM
-DATETIME=$(date "+%m%d%H%M")
-LOGFILE="resnet_green_${DATETIME}_${SEED}.log"
-
-# init
-BENCHMARK=10 INITMLPERF=1 python3 examples/mlperf/model_train.py | tee $LOGFILE
-
-# run
-PARALLEL=0 RUNMLPERF=1 EVAL_START_EPOCH=3 EVAL_FREQ=4 python3 examples/mlperf/model_train.py | tee -a $LOGFILE
@@ -1,50 +0,0 @@
-# 1. Problem
-
-This problem uses the ResNet-50 CNN to do image classification.
-
-## Requirements
-
-Install tinygrad and mlperf-logging from master.
-```
-git clone https://github.com/tinygrad/tinygrad.git
-python3 -m pip install -e ".[mlperf]"
-```
-
-### tinybox_green
-Install the p2p driver per [README](https://github.com/tinygrad/open-gpu-kernel-modules/blob/550.54.15-p2p/README.md)
-This is the default on production tinybox green.
-
-### tinybox_red
-Disable cwsr
-This is the default on production tinybox red.
-```
-sudo vi /etc/modprobe.d/amdgpu.conf
-cat <<EOF > /etc/modprobe.d/amdgpu.conf
-options amdgpu cwsr_enable=0
-EOF
-sudo update-initramfs -u
-sudo reboot
-
-# validate
-sudo cat /sys/module/amdgpu/parameters/cwsr_enable #= 0
-```
-
-# 2. Directions
-
-## Steps to download and verify data
-
-```
-IMGNET_TRAIN=1 python3 extra/datasets/imagenet_download.py
-```
-
-## Steps for one time setup
-
-### tinybox_red
-```
-examples/mlperf/training_submission_v4.0/tinycorp/benchmarks/resnet/implementations/tinybox_red/setup.sh
-```
-
-## Steps to run benchmark
-```
-examples/mlperf/training_submission_v4.0/tinycorp/benchmarks/resnet/implementations/tinybox_red/run_and_time.sh
-```
@@ -1,13 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="resnet"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=1536 EVAL_BS=192
-
-export LAZYCACHE=0 RESET_STEP=0
-
-export TRAIN_BEAM=4 IGNORE_JIT_FIRST_BEAM=1 BEAM_UOPS_MAX=2000 BEAM_UPCAST_MAX=96 BEAM_LOCAL_MAX=1024 BEAM_MIN_PROGRESS=5 BEAM_PADTO=0
-
-export BENCHMARK=10 DEBUG=2
-
-python3 examples/mlperf/model_train.py
@@ -1,15 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="resnet"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=1536 EVAL_BS=192
-
-export LAZYCACHE=0 RESET_STEP=0
-
-export TRAIN_BEAM=4 IGNORE_JIT_FIRST_BEAM=1 BEAM_UOPS_MAX=2000 BEAM_UPCAST_MAX=96 BEAM_LOCAL_MAX=1024 BEAM_MIN_PROGRESS=5 BEAM_PADTO=0
-
-export EVAL_START_EPOCH=3 EVAL_FREQ=4
-
-export WANDB=1 PARALLEL=0
-
-python3 examples/mlperf/model_train.py
@@ -1,23 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="resnet"
-export SUBMISSION_PLATFORM="tinybox_red"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=1536 EVAL_BS=192
-
-export LAZYCACHE=0 RESET_STEP=0
-
-export TRAIN_BEAM=4 IGNORE_JIT_FIRST_BEAM=1 BEAM_UOPS_MAX=2000 BEAM_UPCAST_MAX=96 BEAM_LOCAL_MAX=1024 BEAM_MIN_PROGRESS=5 BEAM_PADTO=0
-
-# pip install -e ".[mlperf]"
-export LOGMLPERF=1
-
-export SEED=$RANDOM
-DATETIME=$(date "+%m%d%H%M")
-LOGFILE="resnet_red_${DATETIME}_${SEED}.log"
-
-# init
-BENCHMARK=10 INITMLPERF=1 python3 examples/mlperf/model_train.py | tee $LOGFILE
-
-# run
-PARALLEL=0 RUNMLPERF=1 EVAL_START_EPOCH=3 EVAL_FREQ=4 python3 examples/mlperf/model_train.py | tee -a $LOGFILE
@@ -1,8 +0,0 @@
-#!/bin/bash
-
-rocm-smi --setprofile compute
-rocm-smi --setmclk 3
-rocm-smi --setperflevel high
-
-# power cap to 350W
-echo "350000000" | sudo tee /sys/class/drm/card{1..6}/device/hwmon/hwmon*/power1_cap
@@ -0,0 +1,87 @@
+:::MLLOG {"namespace": "", "time_ms": 1728516968768, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728516968782, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_green", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728516968782, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728516968782, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728516968782, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728516968917, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728516968917, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728518095273, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728518110874, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728518125400, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728518125401, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728518125401, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728518125401, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728518125401, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728518125401, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728518125402, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728518125402, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728518125402, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728518125402, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728518125402, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728518125402, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728518125402, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728518125402, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728518125403, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728518125403, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728518125403, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728518171154, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728519204577, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728519263743, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728519263744, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.38641827217854635, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.38641827217854635}}
+:::MLLOG {"namespace": "", "time_ms": 1728520267792, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728520321266, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728520321266, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.40444660376272445, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.40444660376272445}}
+:::MLLOG {"namespace": "", "time_ms": 1728521322547, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728521376298, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728521376298, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4533385156548231, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.4533385156548231}}
+:::MLLOG {"namespace": "", "time_ms": 1728522377080, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728522429361, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728522429362, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5267527467952778, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.5267527467952778}}
+:::MLLOG {"namespace": "", "time_ms": 1728523431856, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728523485950, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728523485951, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6413663900499224, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.6413663900499224}}
+:::MLLOG {"namespace": "", "time_ms": 1728524487365, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728524539365, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728524539365, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7015928945715869, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.7015928945715869}}
+:::MLLOG {"namespace": "", "time_ms": 1728525540578, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728525593684, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728525593684, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7095211997458373, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.7095211997458373}}
+:::MLLOG {"namespace": "", "time_ms": 1728526596673, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728526649935, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728526649936, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.712105579231768, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.712105579231768}}
+:::MLLOG {"namespace": "", "time_ms": 1728527652974, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728527704774, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728527704774, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7137153520152179, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7137153520152179}}
+:::MLLOG {"namespace": "", "time_ms": 1728528706518, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728528759227, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728528759227, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7148766237672532, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7148766237672532}}
+:::MLLOG {"namespace": "", "time_ms": 1728529761683, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728529814822, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728529814823, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7160955339258992, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7160955339258992}}
+:::MLLOG {"namespace": "", "time_ms": 1728530822786, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728530876907, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728530876907, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.716297444534931, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.716297444534931}}
+:::MLLOG {"namespace": "", "time_ms": 1728531883583, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728531936703, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728531936703, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7177683101775908, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.7177683101775908}}
+:::MLLOG {"namespace": "", "time_ms": 1728532940983, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728532993183, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728532993183, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.718530326026889, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.718530326026889}}
+:::MLLOG {"namespace": "", "time_ms": 1728534011810, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728534065533, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728534065533, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7185927641985298, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.7185927641985298}}
+:::MLLOG {"namespace": "", "time_ms": 1728535067978, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728535122144, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728535122145, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7192297569276619, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.7192297569276619}}
+:::MLLOG {"namespace": "", "time_ms": 1728536131543, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2549184, "step_num": 38624}}
+:::MLLOG {"namespace": "", "time_ms": 1728536184105, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2549184, "step_num": 38624, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728536184105, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7197693878473032, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2549184, "masked_lm_accuracy": 0.7197693878473032}}
+:::MLLOG {"namespace": "", "time_ms": 1728537200158, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2699136, "step_num": 40896}}
+:::MLLOG {"namespace": "", "time_ms": 1728537253408, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2699136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2699136, "step_num": 40896, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728537253408, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7199979490266993, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2699136, "masked_lm_accuracy": 0.7199979490266993}}
+:::MLLOG {"namespace": "", "time_ms": 1728538262628, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2849088, "step_num": 43168}}
+:::MLLOG {"namespace": "", "time_ms": 1728538316185, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2849088, "step_num": 43168, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728538316186, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7208575420416825, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2849088, "masked_lm_accuracy": 0.7208575420416825}}
+:::MLLOG {"namespace": "", "time_ms": 1728538316186, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 2849088}}
+:::MLLOG {"namespace": "", "time_ms": 1728538316186, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728538316186, "event_type": "POINT_IN_TIME", "key": "seed", "value": 6505, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,99 @@
+:::MLLOG {"namespace": "", "time_ms": 1728538334148, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728538334162, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_green", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728538334162, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728538334162, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728538334162, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728538334302, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728538334302, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728539452588, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728539466234, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728539480889, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728539480890, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728539480890, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728539480890, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728539480890, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728539480890, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728539480890, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728539480891, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728539480891, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728539480891, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728539480891, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728539480891, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728539480891, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728539480891, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728539480891, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728539480892, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728539480892, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728539531181, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728540563757, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728540627488, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728540627488, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.38798193001575504, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.38798193001575504}}
+:::MLLOG {"namespace": "", "time_ms": 1728541627605, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728541684840, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728541684840, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4043695551053306, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.4043695551053306}}
+:::MLLOG {"namespace": "", "time_ms": 1728542682448, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728542739985, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728542739985, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.43848311595381845, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.43848311595381845}}
+:::MLLOG {"namespace": "", "time_ms": 1728543736385, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728543792735, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728543792736, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.519008471802029, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.519008471802029}}
+:::MLLOG {"namespace": "", "time_ms": 1728544788834, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728544846253, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728544846253, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6122590443583112, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.6122590443583112}}
+:::MLLOG {"namespace": "", "time_ms": 1728545843895, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728545901605, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728545901606, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.687927868134545, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.687927868134545}}
+:::MLLOG {"namespace": "", "time_ms": 1728546896899, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728546954045, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728546954045, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7037530967627161, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.7037530967627161}}
+:::MLLOG {"namespace": "", "time_ms": 1728547951266, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728548007399, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728548007400, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7094319296154922, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7094319296154922}}
+:::MLLOG {"namespace": "", "time_ms": 1728549002681, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728549061015, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728549061015, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7113743569225913, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7113743569225913}}
+:::MLLOG {"namespace": "", "time_ms": 1728550055497, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728550112524, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728550112524, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7128702056715427, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7128702056715427}}
+:::MLLOG {"namespace": "", "time_ms": 1728551106231, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728551163221, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728551163221, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7141216593941458, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7141216593941458}}
+:::MLLOG {"namespace": "", "time_ms": 1728552166923, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728552223146, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728552223147, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7146758918069978, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.7146758918069978}}
+:::MLLOG {"namespace": "", "time_ms": 1728553218362, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728553275684, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728553275684, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7159770885197503, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.7159770885197503}}
+:::MLLOG {"namespace": "", "time_ms": 1728554271310, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728554328503, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728554328503, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7166727805252052, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7166727805252052}}
+:::MLLOG {"namespace": "", "time_ms": 1728555323857, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728555381026, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728555381026, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7173972926457342, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.7173972926457342}}
+:::MLLOG {"namespace": "", "time_ms": 1728556388293, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728556444413, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728556444413, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7175414781407389, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.7175414781407389}}
+:::MLLOG {"namespace": "", "time_ms": 1728557446881, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2549184, "step_num": 38624}}
+:::MLLOG {"namespace": "", "time_ms": 1728557503528, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2549184, "step_num": 38624, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728557503528, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7179403612909735, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2549184, "masked_lm_accuracy": 0.7179403612909735}}
+:::MLLOG {"namespace": "", "time_ms": 1728558506474, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2699136, "step_num": 40896}}
+:::MLLOG {"namespace": "", "time_ms": 1728558562665, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2699136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2699136, "step_num": 40896, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728558562665, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7183986956585505, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2699136, "masked_lm_accuracy": 0.7183986956585505}}
+:::MLLOG {"namespace": "", "time_ms": 1728559573313, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2849088, "step_num": 43168}}
+:::MLLOG {"namespace": "", "time_ms": 1728559631801, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2849088, "step_num": 43168, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728559631801, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7190847733311119, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2849088, "masked_lm_accuracy": 0.7190847733311119}}
+:::MLLOG {"namespace": "", "time_ms": 1728560634330, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2999040, "step_num": 45440}}
+:::MLLOG {"namespace": "", "time_ms": 1728560690618, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2999040, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2999040, "step_num": 45440, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728560690618, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7193026458280274, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2999040, "masked_lm_accuracy": 0.7193026458280274}}
+:::MLLOG {"namespace": "", "time_ms": 1728561686592, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3148992, "step_num": 47712}}
+:::MLLOG {"namespace": "", "time_ms": 1728561743999, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3148992, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3148992, "step_num": 47712, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728561743999, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7195320777644207, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3148992, "masked_lm_accuracy": 0.7195320777644207}}
+:::MLLOG {"namespace": "", "time_ms": 1728562740953, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3298944, "step_num": 49984}}
+:::MLLOG {"namespace": "", "time_ms": 1728562797084, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3298944, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3298944, "step_num": 49984, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728562797084, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7198491232010441, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3298944, "masked_lm_accuracy": 0.7198491232010441}}
+:::MLLOG {"namespace": "", "time_ms": 1728563794843, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3448896, "step_num": 52256}}
+:::MLLOG {"namespace": "", "time_ms": 1728563851217, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3448896, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3448896, "step_num": 52256, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728563851218, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7201704975009752, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3448896, "masked_lm_accuracy": 0.7201704975009752}}
+:::MLLOG {"namespace": "", "time_ms": 1728563851218, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 3448896, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 3448896}}
+:::MLLOG {"namespace": "", "time_ms": 1728563851218, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728563851218, "event_type": "POINT_IN_TIME", "key": "seed", "value": 20151, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,100 @@
+:::MLLOG {"namespace": "", "time_ms": 1728563867610, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728563867623, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_green", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728563867623, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728563867623, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728563867623, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728563867760, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728563867761, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728564959716, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728564973303, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728564988006, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728564988006, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728564988006, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728564988006, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728564988006, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728564988007, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728564988007, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728564988007, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728564988007, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728564988007, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728564988007, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728564988007, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728564988008, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728564988008, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728564988008, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728564988008, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728564988008, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728565046616, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728566082579, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728566146228, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728566146228, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.3875764702688668, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.3875764702688668}}
+:::MLLOG {"namespace": "", "time_ms": 1728567148838, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728567205854, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728567205855, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4064989212786429, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.4064989212786429}}
+:::MLLOG {"namespace": "", "time_ms": 1728568204176, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728568261886, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728568261886, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4445486674068499, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.4445486674068499}}
+:::MLLOG {"namespace": "", "time_ms": 1728569260373, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728569317827, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728569317827, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5039052128362741, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.5039052128362741}}
+:::MLLOG {"namespace": "", "time_ms": 1728570316014, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728570373692, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728570373692, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5860576768012982, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.5860576768012982}}
+:::MLLOG {"namespace": "", "time_ms": 1728571372081, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728571429048, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728571429048, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6553790274678981, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.6553790274678981}}
+:::MLLOG {"namespace": "", "time_ms": 1728572429036, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728572485646, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728572485646, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.695299510013292, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.695299510013292}}
+:::MLLOG {"namespace": "", "time_ms": 1728573485353, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728573541985, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728573541985, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7060198057319994, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7060198057319994}}
+:::MLLOG {"namespace": "", "time_ms": 1728574539420, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728574596892, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728574596892, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7096440969074137, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7096440969074137}}
+:::MLLOG {"namespace": "", "time_ms": 1728575594573, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728575651826, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728575651827, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7117632173033053, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7117632173033053}}
+:::MLLOG {"namespace": "", "time_ms": 1728576649028, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728576706272, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728576706273, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7133303043652096, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7133303043652096}}
+:::MLLOG {"namespace": "", "time_ms": 1728577704079, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728577761356, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728577761356, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7137490666835123, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.7137490666835123}}
+:::MLLOG {"namespace": "", "time_ms": 1728578758647, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728578815837, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728578815837, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7155028079610137, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.7155028079610137}}
+:::MLLOG {"namespace": "", "time_ms": 1728579812964, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728579870775, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728579870776, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7161324361018528, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7161324361018528}}
+:::MLLOG {"namespace": "", "time_ms": 1728580866942, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728580924299, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728580924299, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7167852242763842, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.7167852242763842}}
+:::MLLOG {"namespace": "", "time_ms": 1728581922472, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728581979021, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728581979022, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7176919813705335, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.7176919813705335}}
+:::MLLOG {"namespace": "", "time_ms": 1728582983990, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2549184, "step_num": 38624}}
+:::MLLOG {"namespace": "", "time_ms": 1728583040463, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2549184, "step_num": 38624, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728583040464, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7181049583197069, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2549184, "masked_lm_accuracy": 0.7181049583197069}}
+:::MLLOG {"namespace": "", "time_ms": 1728584044016, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2699136, "step_num": 40896}}
+:::MLLOG {"namespace": "", "time_ms": 1728584101869, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2699136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2699136, "step_num": 40896, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728584101869, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7182932556855443, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2699136, "masked_lm_accuracy": 0.7182932556855443}}
+:::MLLOG {"namespace": "", "time_ms": 1728585118787, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2849088, "step_num": 43168}}
+:::MLLOG {"namespace": "", "time_ms": 1728585176571, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2849088, "step_num": 43168, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728585176571, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7189288123968338, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2849088, "masked_lm_accuracy": 0.7189288123968338}}
+:::MLLOG {"namespace": "", "time_ms": 1728586176081, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2999040, "step_num": 45440}}
+:::MLLOG {"namespace": "", "time_ms": 1728586233843, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2999040, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2999040, "step_num": 45440, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728586233843, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7190464254475384, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2999040, "masked_lm_accuracy": 0.7190464254475384}}
+:::MLLOG {"namespace": "", "time_ms": 1728587233422, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3148992, "step_num": 47712}}
+:::MLLOG {"namespace": "", "time_ms": 1728587290110, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3148992, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3148992, "step_num": 47712, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728587290111, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7189439674587971, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3148992, "masked_lm_accuracy": 0.7189439674587971}}
+:::MLLOG {"namespace": "", "time_ms": 1728588298920, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3298944, "step_num": 49984}}
+:::MLLOG {"namespace": "", "time_ms": 1728588356857, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3298944, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3298944, "step_num": 49984, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728588356858, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7195348424974429, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3298944, "masked_lm_accuracy": 0.7195348424974429}}
+:::MLLOG {"namespace": "", "time_ms": 1728589362222, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3448896, "step_num": 52256}}
+:::MLLOG {"namespace": "", "time_ms": 1728589419360, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3448896, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3448896, "step_num": 52256, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728589419360, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7198157412389402, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3448896, "masked_lm_accuracy": 0.7198157412389402}}
+:::MLLOG {"namespace": "", "time_ms": 1728590419216, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3598848, "step_num": 54528}}
+:::MLLOG {"namespace": "", "time_ms": 1728590477169, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3598848, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3598848, "step_num": 54528, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728590477170, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7196774663650568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3598848, "masked_lm_accuracy": 0.7196774663650568}}
+:::MLLOG {"namespace": "", "time_ms": 1728590477170, "event_type": "POINT_IN_TIME", "key": "seed", "value": 15936, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,78 @@
+:::MLLOG {"namespace": "", "time_ms": 1728590796935, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728590796948, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_green", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728590796948, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728590796949, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728590796949, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728590797097, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728590797098, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728591915361, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728591929057, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943823, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943823, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943824, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943824, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943824, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943824, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943824, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943824, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943824, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943825, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943825, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943825, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943825, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943825, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943825, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943825, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943826, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728591989241, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728593032285, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728593096400, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728593096400, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.38762320266130373, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.38762320266130373}}
+:::MLLOG {"namespace": "", "time_ms": 1728594107968, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728594165957, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728594165958, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.41101139415576204, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.41101139415576204}}
+:::MLLOG {"namespace": "", "time_ms": 1728595175795, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728595232491, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728595232492, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4496020218106037, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.4496020218106037}}
+:::MLLOG {"namespace": "", "time_ms": 1728596242699, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728596299316, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728596299317, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5736796348911599, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.5736796348911599}}
+:::MLLOG {"namespace": "", "time_ms": 1728597307215, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728597365230, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728597365230, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6932630223218166, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.6932630223218166}}
+:::MLLOG {"namespace": "", "time_ms": 1728598375878, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728598433374, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728598433374, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7074951962503617, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.7074951962503617}}
+:::MLLOG {"namespace": "", "time_ms": 1728599442119, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728599498818, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728599498818, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7122511012366809, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.7122511012366809}}
+:::MLLOG {"namespace": "", "time_ms": 1728600507693, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728600566544, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728600566545, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7142625644525941, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7142625644525941}}
+:::MLLOG {"namespace": "", "time_ms": 1728601572961, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728601630426, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728601630427, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7150282886618973, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7150282886618973}}
+:::MLLOG {"namespace": "", "time_ms": 1728602641519, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728602697877, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728602697878, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7163425168378953, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7163425168378953}}
+:::MLLOG {"namespace": "", "time_ms": 1728603708872, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728603765541, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728603765541, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7169951724305293, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7169951724305293}}
+:::MLLOG {"namespace": "", "time_ms": 1728604783436, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728604840983, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728604840983, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7174601919220533, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.7174601919220533}}
+:::MLLOG {"namespace": "", "time_ms": 1728605853878, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728605911397, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728605911397, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.718169204153268, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.718169204153268}}
+:::MLLOG {"namespace": "", "time_ms": 1728606921863, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728606978402, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728606978403, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7192717549253096, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7192717549253096}}
+:::MLLOG {"namespace": "", "time_ms": 1728607995812, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728608053682, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728608053682, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7194449275499629, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.7194449275499629}}
+:::MLLOG {"namespace": "", "time_ms": 1728609084329, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728609142724, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728609142725, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7200488402375792, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.7200488402375792}}
+:::MLLOG {"namespace": "", "time_ms": 1728609142725, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 2399232}}
+:::MLLOG {"namespace": "", "time_ms": 1728609142725, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728609142725, "event_type": "POINT_IN_TIME", "key": "seed", "value": 20762, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,96 @@
+:::MLLOG {"namespace": "", "time_ms": 1728609159110, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728609159123, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_green", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728609159123, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728609159124, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728609159124, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728609159270, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728609159271, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728610267217, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728610281092, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728610295781, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728610295781, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728610295781, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728610295782, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728610295782, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728610295782, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728610295782, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728610295782, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728610295782, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728610295782, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728610295782, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728610295783, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728610295783, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728610295783, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728610295783, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728610295783, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728610295783, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728610347873, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728611397047, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728611461353, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728611461353, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.38775324024121494, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.38775324024121494}}
+:::MLLOG {"namespace": "", "time_ms": 1728612481307, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728612538833, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728612538833, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4005705962727437, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.4005705962727437}}
+:::MLLOG {"namespace": "", "time_ms": 1728613556463, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728613613480, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728613613480, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4705434759434069, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.4705434759434069}}
+:::MLLOG {"namespace": "", "time_ms": 1728614631311, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728614687851, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728614687851, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5429764502622013, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.5429764502622013}}
+:::MLLOG {"namespace": "", "time_ms": 1728615706690, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728615763381, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728615763381, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6442740782693109, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.6442740782693109}}
+:::MLLOG {"namespace": "", "time_ms": 1728616781291, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728616838087, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728616838087, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6980952795351345, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.6980952795351345}}
+:::MLLOG {"namespace": "", "time_ms": 1728617855821, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728617912283, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728617912283, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.707689059052413, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.707689059052413}}
+:::MLLOG {"namespace": "", "time_ms": 1728618929847, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728618986661, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728618986661, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7109381170707616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7109381170707616}}
+:::MLLOG {"namespace": "", "time_ms": 1728620004644, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728620061231, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728620061231, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.712553325461712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.712553325461712}}
+:::MLLOG {"namespace": "", "time_ms": 1728621080635, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728621138171, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728621138171, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.714370012497859, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.714370012497859}}
+:::MLLOG {"namespace": "", "time_ms": 1728622165561, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728622221929, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728622221929, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7153810627029982, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7153810627029982}}
+:::MLLOG {"namespace": "", "time_ms": 1728623242362, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728623299587, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728623299588, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7157506428582981, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.7157506428582981}}
+:::MLLOG {"namespace": "", "time_ms": 1728624319084, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728624376686, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728624376687, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.716899270201845, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.716899270201845}}
+:::MLLOG {"namespace": "", "time_ms": 1728625402109, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728625460126, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728625460126, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7170993249861152, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7170993249861152}}
+:::MLLOG {"namespace": "", "time_ms": 1728626480567, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728626539447, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728626539448, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7173656942820077, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.7173656942820077}}
+:::MLLOG {"namespace": "", "time_ms": 1728627566869, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728627623861, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728627623862, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7182765536917565, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.7182765536917565}}
+:::MLLOG {"namespace": "", "time_ms": 1728628643742, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2549184, "step_num": 38624}}
+:::MLLOG {"namespace": "", "time_ms": 1728628701681, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2549184, "step_num": 38624, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728628701681, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7185246580435118, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2549184, "masked_lm_accuracy": 0.7185246580435118}}
+:::MLLOG {"namespace": "", "time_ms": 1728629728019, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2699136, "step_num": 40896}}
+:::MLLOG {"namespace": "", "time_ms": 1728629785903, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2699136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2699136, "step_num": 40896, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728629785903, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7187082183954597, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2699136, "masked_lm_accuracy": 0.7187082183954597}}
+:::MLLOG {"namespace": "", "time_ms": 1728630806207, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2849088, "step_num": 43168}}
+:::MLLOG {"namespace": "", "time_ms": 1728630863277, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2849088, "step_num": 43168, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728630863277, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7192723798623111, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2849088, "masked_lm_accuracy": 0.7192723798623111}}
+:::MLLOG {"namespace": "", "time_ms": 1728631896497, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2999040, "step_num": 45440}}
+:::MLLOG {"namespace": "", "time_ms": 1728631954615, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2999040, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2999040, "step_num": 45440, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728631954616, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7195008011299047, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2999040, "masked_lm_accuracy": 0.7195008011299047}}
+:::MLLOG {"namespace": "", "time_ms": 1728632984222, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3148992, "step_num": 47712}}
+:::MLLOG {"namespace": "", "time_ms": 1728633041346, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3148992, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3148992, "step_num": 47712, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728633041346, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7195586102339202, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3148992, "masked_lm_accuracy": 0.7195586102339202}}
+:::MLLOG {"namespace": "", "time_ms": 1728634068369, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3298944, "step_num": 49984}}
+:::MLLOG {"namespace": "", "time_ms": 1728634127804, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3298944, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3298944, "step_num": 49984, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728634127805, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7202416826810534, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3298944, "masked_lm_accuracy": 0.7202416826810534}}
+:::MLLOG {"namespace": "", "time_ms": 1728634127805, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 3298944, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 3298944}}
+:::MLLOG {"namespace": "", "time_ms": 1728634127805, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728634127805, "event_type": "POINT_IN_TIME", "key": "seed", "value": 219, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,93 @@
+:::MLLOG {"namespace": "", "time_ms": 1728524965239, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728524965253, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_green", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728524965253, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728524965253, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728524965253, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728524965392, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728524965393, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728526062933, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728526077625, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728526092468, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728526092468, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728526092468, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728526092468, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728526092469, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728526092469, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728526092469, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728526092469, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728526092469, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728526092469, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728526092469, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728526092470, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728526092470, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728526092470, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728526092470, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728526092470, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728526092470, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728526138852, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728527173955, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728527237057, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728527237057, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.3877335678748049, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.3877335678748049}}
+:::MLLOG {"namespace": "", "time_ms": 1728528242671, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728528300179, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728528300179, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.40263391588716785, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.40263391588716785}}
+:::MLLOG {"namespace": "", "time_ms": 1728529303573, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728529359959, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728529359959, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.42964756483436706, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.42964756483436706}}
+:::MLLOG {"namespace": "", "time_ms": 1728530364689, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728530422126, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728530422126, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4922114010334873, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.4922114010334873}}
+:::MLLOG {"namespace": "", "time_ms": 1728531424135, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728531481285, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728531481286, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5822947872707639, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.5822947872707639}}
+:::MLLOG {"namespace": "", "time_ms": 1728532482693, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728532539641, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728532539641, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6813526992749224, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.6813526992749224}}
+:::MLLOG {"namespace": "", "time_ms": 1728533542012, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728533599055, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728533599056, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7051245396052854, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.7051245396052854}}
+:::MLLOG {"namespace": "", "time_ms": 1728534601465, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728534657337, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728534657337, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7107940144930761, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7107940144930761}}
+:::MLLOG {"namespace": "", "time_ms": 1728535659894, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728535716827, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728535716827, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7130225877169728, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7130225877169728}}
+:::MLLOG {"namespace": "", "time_ms": 1728536719704, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728536776465, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728536776465, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7149506402573474, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7149506402573474}}
+:::MLLOG {"namespace": "", "time_ms": 1728537786290, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728537844483, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728537844483, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7156235970418183, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7156235970418183}}
+:::MLLOG {"namespace": "", "time_ms": 1728538854316, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728538910323, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728538910323, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.716550941694691, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.716550941694691}}
+:::MLLOG {"namespace": "", "time_ms": 1728539924500, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728539982702, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728539982702, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7173966967065533, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.7173966967065533}}
+:::MLLOG {"namespace": "", "time_ms": 1728540985244, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728541042328, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728541042329, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7179239220295971, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7179239220295971}}
+:::MLLOG {"namespace": "", "time_ms": 1728542045079, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728542102249, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728542102249, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7180614758195746, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.7180614758195746}}
+:::MLLOG {"namespace": "", "time_ms": 1728543106186, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728543163150, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728543163150, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7188610007967431, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.7188610007967431}}
+:::MLLOG {"namespace": "", "time_ms": 1728544166236, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2549184, "step_num": 38624}}
+:::MLLOG {"namespace": "", "time_ms": 1728544222285, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2549184, "step_num": 38624, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728544222285, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7190189941528677, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2549184, "masked_lm_accuracy": 0.7190189941528677}}
+:::MLLOG {"namespace": "", "time_ms": 1728545224291, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2699136, "step_num": 40896}}
+:::MLLOG {"namespace": "", "time_ms": 1728545280273, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2699136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2699136, "step_num": 40896, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728545280273, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7189939534442469, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2699136, "masked_lm_accuracy": 0.7189939534442469}}
+:::MLLOG {"namespace": "", "time_ms": 1728546290997, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2849088, "step_num": 43168}}
+:::MLLOG {"namespace": "", "time_ms": 1728546348255, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2849088, "step_num": 43168, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728546348255, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7197046605641069, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2849088, "masked_lm_accuracy": 0.7197046605641069}}
+:::MLLOG {"namespace": "", "time_ms": 1728547370392, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2999040, "step_num": 45440}}
+:::MLLOG {"namespace": "", "time_ms": 1728547427703, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2999040, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2999040, "step_num": 45440, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728547427704, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7198148297920296, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2999040, "masked_lm_accuracy": 0.7198148297920296}}
+:::MLLOG {"namespace": "", "time_ms": 1728548433236, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3148992, "step_num": 47712}}
+:::MLLOG {"namespace": "", "time_ms": 1728548489162, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3148992, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3148992, "step_num": 47712, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728548489163, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7201927967892483, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3148992, "masked_lm_accuracy": 0.7201927967892483}}
+:::MLLOG {"namespace": "", "time_ms": 1728548489163, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 3148992, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 3148992}}
+:::MLLOG {"namespace": "", "time_ms": 1728548489163, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728548489163, "event_type": "POINT_IN_TIME", "key": "seed", "value": 28210, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,87 @@
+:::MLLOG {"namespace": "", "time_ms": 1728548506444, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728548506457, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_green", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728548506458, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728548506458, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728548506458, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728548506735, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728548506735, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728549621641, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728549635270, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728549652872, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728549652872, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728549652872, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728549652872, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728549652872, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728549652873, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728549652873, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728549652873, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728549652873, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728549652873, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728549652873, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728549652873, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728549652873, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728549652873, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728549652874, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728549652874, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728549652874, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728549705049, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728550749936, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728550814006, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728550814007, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.3879261584913366, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.3879261584913366}}
+:::MLLOG {"namespace": "", "time_ms": 1728551828200, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728551884657, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728551884657, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.40485212616242544, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.40485212616242544}}
+:::MLLOG {"namespace": "", "time_ms": 1728552894731, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728552950982, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728552950982, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.44991275376735795, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.44991275376735795}}
+:::MLLOG {"namespace": "", "time_ms": 1728553961756, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728554019767, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728554019768, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5169473737210089, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.5169473737210089}}
+:::MLLOG {"namespace": "", "time_ms": 1728555028196, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728555084535, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728555084536, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6089998500284207, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.6089998500284207}}
+:::MLLOG {"namespace": "", "time_ms": 1728556093244, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728556150861, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728556150861, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6900064689568152, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.6900064689568152}}
+:::MLLOG {"namespace": "", "time_ms": 1728557159270, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728557216649, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728557216649, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7058100798706416, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.7058100798706416}}
+:::MLLOG {"namespace": "", "time_ms": 1728558223895, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728558281962, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728558281962, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7103145412148726, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7103145412148726}}
+:::MLLOG {"namespace": "", "time_ms": 1728559289692, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728559346801, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728559346801, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7119302401016341, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7119302401016341}}
+:::MLLOG {"namespace": "", "time_ms": 1728560353303, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728560410326, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728560410327, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7139668108176956, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7139668108176956}}
+:::MLLOG {"namespace": "", "time_ms": 1728561418512, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728561474810, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728561474810, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.715211286184383, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.715211286184383}}
+:::MLLOG {"namespace": "", "time_ms": 1728562483778, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728562541169, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728562541169, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7157247102968551, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.7157247102968551}}
+:::MLLOG {"namespace": "", "time_ms": 1728563551560, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728563609899, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728563609899, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.717246541128805, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.717246541128805}}
+:::MLLOG {"namespace": "", "time_ms": 1728564615944, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728564673046, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728564673046, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7174253223824801, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7174253223824801}}
+:::MLLOG {"namespace": "", "time_ms": 1728565680391, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728565737965, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728565737965, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7175353851658753, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.7175353851658753}}
+:::MLLOG {"namespace": "", "time_ms": 1728566749670, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728566807129, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728566807130, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7185968662280842, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.7185968662280842}}
+:::MLLOG {"namespace": "", "time_ms": 1728567827685, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2549184, "step_num": 38624}}
+:::MLLOG {"namespace": "", "time_ms": 1728567883857, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2549184, "step_num": 38624, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728567883857, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7189708248755141, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2549184, "masked_lm_accuracy": 0.7189708248755141}}
+:::MLLOG {"namespace": "", "time_ms": 1728568903914, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2699136, "step_num": 40896}}
+:::MLLOG {"namespace": "", "time_ms": 1728568961416, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2699136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2699136, "step_num": 40896, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728568961416, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7191977317131559, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2699136, "masked_lm_accuracy": 0.7191977317131559}}
+:::MLLOG {"namespace": "", "time_ms": 1728569970441, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2849088, "step_num": 43168}}
+:::MLLOG {"namespace": "", "time_ms": 1728570027956, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2849088, "step_num": 43168, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728570027956, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7201856829552287, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2849088, "masked_lm_accuracy": 0.7201856829552287}}
+:::MLLOG {"namespace": "", "time_ms": 1728570027957, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 2849088}}
+:::MLLOG {"namespace": "", "time_ms": 1728570027957, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728570027957, "event_type": "POINT_IN_TIME", "key": "seed", "value": 10448, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,96 @@
+:::MLLOG {"namespace": "", "time_ms": 1728570044042, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728570044056, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_green", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728570044056, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728570044056, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728570044056, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728570044375, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728570044375, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728571155314, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728571169188, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728571185077, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728571185077, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728571185077, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728571185077, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728571185077, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728571185078, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728571185078, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728571185078, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728571185078, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728571185078, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728571185078, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728571185078, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728571185079, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728571185079, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728571185079, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728571185079, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728571185079, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728571232077, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728572260932, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728572324272, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728572324272, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.3880780045687258, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.3880780045687258}}
+:::MLLOG {"namespace": "", "time_ms": 1728573321862, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728573379098, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728573379099, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4047361867257629, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.4047361867257629}}
+:::MLLOG {"namespace": "", "time_ms": 1728574372317, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728574430334, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728574430334, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4497806984254108, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.4497806984254108}}
+:::MLLOG {"namespace": "", "time_ms": 1728575423648, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728575480052, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728575480052, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5134892886673252, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.5134892886673252}}
+:::MLLOG {"namespace": "", "time_ms": 1728576473686, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728576529490, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728576529490, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6072650871808899, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.6072650871808899}}
+:::MLLOG {"namespace": "", "time_ms": 1728577521455, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728577578395, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728577578395, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6880894238246581, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.6880894238246581}}
+:::MLLOG {"namespace": "", "time_ms": 1728578569253, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728578627203, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728578627203, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7046831176176569, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.7046831176176569}}
+:::MLLOG {"namespace": "", "time_ms": 1728579618293, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728579675086, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728579675087, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7089016024552924, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7089016024552924}}
+:::MLLOG {"namespace": "", "time_ms": 1728580666335, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728580724803, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728580724804, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7109720907171257, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7109720907171257}}
+:::MLLOG {"namespace": "", "time_ms": 1728581716985, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728581773028, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728581773028, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7131220351884519, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7131220351884519}}
+:::MLLOG {"namespace": "", "time_ms": 1728582764878, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728582822486, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728582822487, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7144350144916047, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7144350144916047}}
+:::MLLOG {"namespace": "", "time_ms": 1728583814857, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728583872551, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728583872551, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7149028261407235, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.7149028261407235}}
+:::MLLOG {"namespace": "", "time_ms": 1728584862980, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728584919984, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728584919984, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7159077662798052, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.7159077662798052}}
+:::MLLOG {"namespace": "", "time_ms": 1728585920409, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728585976497, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728585976497, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7165625739683987, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7165625739683987}}
+:::MLLOG {"namespace": "", "time_ms": 1728586977974, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728587034924, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728587034924, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.716636929886743, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.716636929886743}}
+:::MLLOG {"namespace": "", "time_ms": 1728588046632, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728588103542, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728588103543, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7177939713180029, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.7177939713180029}}
+:::MLLOG {"namespace": "", "time_ms": 1728589095819, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2549184, "step_num": 38624}}
+:::MLLOG {"namespace": "", "time_ms": 1728589153039, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2549184, "step_num": 38624, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728589153040, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.718201180382553, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2549184, "masked_lm_accuracy": 0.718201180382553}}
+:::MLLOG {"namespace": "", "time_ms": 1728590145828, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2699136, "step_num": 40896}}
+:::MLLOG {"namespace": "", "time_ms": 1728590203087, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2699136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2699136, "step_num": 40896, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728590203087, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7187695578679255, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2699136, "masked_lm_accuracy": 0.7187695578679255}}
+:::MLLOG {"namespace": "", "time_ms": 1728591195213, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2849088, "step_num": 43168}}
+:::MLLOG {"namespace": "", "time_ms": 1728591252415, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2849088, "step_num": 43168, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728591252415, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7192902017941215, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2849088, "masked_lm_accuracy": 0.7192902017941215}}
+:::MLLOG {"namespace": "", "time_ms": 1728592251653, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2999040, "step_num": 45440}}
+:::MLLOG {"namespace": "", "time_ms": 1728592309614, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2999040, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2999040, "step_num": 45440, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728592309614, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7194759279721928, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2999040, "masked_lm_accuracy": 0.7194759279721928}}
+:::MLLOG {"namespace": "", "time_ms": 1728593301651, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3148992, "step_num": 47712}}
+:::MLLOG {"namespace": "", "time_ms": 1728593358684, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3148992, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3148992, "step_num": 47712, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728593358684, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7194070329763393, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3148992, "masked_lm_accuracy": 0.7194070329763393}}
+:::MLLOG {"namespace": "", "time_ms": 1728594357831, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3298944, "step_num": 49984}}
+:::MLLOG {"namespace": "", "time_ms": 1728594414864, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3298944, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3298944, "step_num": 49984, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728594414864, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7200078809769052, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3298944, "masked_lm_accuracy": 0.7200078809769052}}
+:::MLLOG {"namespace": "", "time_ms": 1728594414864, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 3298944, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 3298944}}
+:::MLLOG {"namespace": "", "time_ms": 1728594414864, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728594414864, "event_type": "POINT_IN_TIME", "key": "seed", "value": 10752, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,78 @@
+:::MLLOG {"namespace": "", "time_ms": 1728590796935, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728590796948, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_green", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728590796948, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728590796949, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728590796949, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728590797097, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728590797098, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728591915361, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728591929057, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943823, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943823, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943824, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943824, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943824, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943824, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943824, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943824, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943824, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943825, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943825, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943825, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943825, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943825, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943825, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943825, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728591943826, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728591989241, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728593032285, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728593096400, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728593096400, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.38762320266130373, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.38762320266130373}}
+:::MLLOG {"namespace": "", "time_ms": 1728594107968, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728594165957, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728594165958, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.41101139415576204, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.41101139415576204}}
+:::MLLOG {"namespace": "", "time_ms": 1728595175795, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728595232491, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728595232492, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4496020218106037, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.4496020218106037}}
+:::MLLOG {"namespace": "", "time_ms": 1728596242699, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728596299316, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728596299317, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5736796348911599, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.5736796348911599}}
+:::MLLOG {"namespace": "", "time_ms": 1728597307215, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728597365230, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728597365230, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6932630223218166, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.6932630223218166}}
+:::MLLOG {"namespace": "", "time_ms": 1728598375878, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728598433374, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728598433374, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7074951962503617, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.7074951962503617}}
+:::MLLOG {"namespace": "", "time_ms": 1728599442119, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728599498818, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728599498818, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7122511012366809, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.7122511012366809}}
+:::MLLOG {"namespace": "", "time_ms": 1728600507693, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728600566544, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728600566545, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7142625644525941, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7142625644525941}}
+:::MLLOG {"namespace": "", "time_ms": 1728601572961, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728601630426, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728601630427, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7150282886618973, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7150282886618973}}
+:::MLLOG {"namespace": "", "time_ms": 1728602641519, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728602697877, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728602697878, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7163425168378953, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7163425168378953}}
+:::MLLOG {"namespace": "", "time_ms": 1728603708872, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728603765541, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728603765541, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7169951724305293, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7169951724305293}}
+:::MLLOG {"namespace": "", "time_ms": 1728604783436, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728604840983, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728604840983, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7174601919220533, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.7174601919220533}}
+:::MLLOG {"namespace": "", "time_ms": 1728605853878, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728605911397, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728605911397, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.718169204153268, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.718169204153268}}
+:::MLLOG {"namespace": "", "time_ms": 1728606921863, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728606978402, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728606978403, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7192717549253096, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7192717549253096}}
+:::MLLOG {"namespace": "", "time_ms": 1728607995812, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728608053682, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728608053682, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7194449275499629, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.7194449275499629}}
+:::MLLOG {"namespace": "", "time_ms": 1728609084329, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728609142724, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728609142725, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7200488402375792, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.7200488402375792}}
+:::MLLOG {"namespace": "", "time_ms": 1728609142725, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 2399232}}
+:::MLLOG {"namespace": "", "time_ms": 1728609142725, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728609142725, "event_type": "POINT_IN_TIME", "key": "seed", "value": 22978, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,90 @@
+:::MLLOG {"namespace": "", "time_ms": 1728618801034, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728618801047, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_green", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728618801048, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728618801048, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728618801048, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728618801209, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728618801209, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728619929327, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728619943087, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728619957912, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728619957913, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728619957913, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728619957913, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728619957913, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728619957913, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728619957913, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728619957913, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728619957914, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728619957914, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728619957914, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728619957914, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728619957914, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728619957914, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728619957914, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728619957914, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728619957915, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728620007681, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728621045859, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728621108773, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728621108773, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.3865617481387155, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.3865617481387155}}
+:::MLLOG {"namespace": "", "time_ms": 1728622114605, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728622171150, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728622171151, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.40262117073717557, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.40262117073717557}}
+:::MLLOG {"namespace": "", "time_ms": 1728623174052, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728623230873, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728623230873, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4523024549724531, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.4523024549724531}}
+:::MLLOG {"namespace": "", "time_ms": 1728624233281, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728624289615, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728624289616, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5221301693185476, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.5221301693185476}}
+:::MLLOG {"namespace": "", "time_ms": 1728625291673, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728625349229, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728625349230, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6420224388345102, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.6420224388345102}}
+:::MLLOG {"namespace": "", "time_ms": 1728626351043, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728626408667, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728626408667, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.699528387405233, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.699528387405233}}
+:::MLLOG {"namespace": "", "time_ms": 1728627410463, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728627466811, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728627466811, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7079550924241078, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.7079550924241078}}
+:::MLLOG {"namespace": "", "time_ms": 1728628468561, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728628525964, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728628525965, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7118257526945195, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7118257526945195}}
+:::MLLOG {"namespace": "", "time_ms": 1728629528193, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728629584413, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728629584413, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7130748185842568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7130748185842568}}
+:::MLLOG {"namespace": "", "time_ms": 1728630587761, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728630645519, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728630645520, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.714902451421661, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.714902451421661}}
+:::MLLOG {"namespace": "", "time_ms": 1728631646942, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728631704444, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728631704445, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7154911232337883, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7154911232337883}}
+:::MLLOG {"namespace": "", "time_ms": 1728632715855, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728632772059, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728632772059, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7161726189646523, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.7161726189646523}}
+:::MLLOG {"namespace": "", "time_ms": 1728633776195, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728633832753, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728633832753, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7168811186745844, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.7168811186745844}}
+:::MLLOG {"namespace": "", "time_ms": 1728634841112, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728634897563, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728634897563, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7172674153762159, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7172674153762159}}
+:::MLLOG {"namespace": "", "time_ms": 1728635899496, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728635956963, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728635956963, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.717885359040119, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.717885359040119}}
+:::MLLOG {"namespace": "", "time_ms": 1728636964772, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728637021152, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728637021152, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.718571519987556, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.718571519987556}}
+:::MLLOG {"namespace": "", "time_ms": 1728638039702, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2549184, "step_num": 38624}}
+:::MLLOG {"namespace": "", "time_ms": 1728638096247, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2549184, "step_num": 38624, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728638096247, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7190522497187993, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2549184, "masked_lm_accuracy": 0.7190522497187993}}
+:::MLLOG {"namespace": "", "time_ms": 1728639099786, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2699136, "step_num": 40896}}
+:::MLLOG {"namespace": "", "time_ms": 1728639156384, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2699136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2699136, "step_num": 40896, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728639156384, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7190969545610951, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2699136, "masked_lm_accuracy": 0.7190969545610951}}
+:::MLLOG {"namespace": "", "time_ms": 1728640172070, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2849088, "step_num": 43168}}
+:::MLLOG {"namespace": "", "time_ms": 1728640229709, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2849088, "step_num": 43168, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728640229709, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7197579530877272, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2849088, "masked_lm_accuracy": 0.7197579530877272}}
+:::MLLOG {"namespace": "", "time_ms": 1728641233638, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2999040, "step_num": 45440}}
+:::MLLOG {"namespace": "", "time_ms": 1728641290427, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2999040, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2999040, "step_num": 45440, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728641290427, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7200084952587272, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2999040, "masked_lm_accuracy": 0.7200084952587272}}
+:::MLLOG {"namespace": "", "time_ms": 1728641290427, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 2999040, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 2999040}}
+:::MLLOG {"namespace": "", "time_ms": 1728641290428, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728641290428, "event_type": "POINT_IN_TIME", "key": "seed", "value": 9634, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,93 @@
+:::MLLOG {"namespace": "", "time_ms": 1728516945293, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728516945306, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_red", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728516945306, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728516945307, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728516945307, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728516945490, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728516945491, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728518502722, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728518514008, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728518528257, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728518528257, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728518528258, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728518528258, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728518528258, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728518528258, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728518528258, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728518528258, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728518528258, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728518528259, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728518528259, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728518528259, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728518528259, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728518528259, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728518528259, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728518528259, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728518528259, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728518578008, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728519789747, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728519848812, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728519848813, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.38834677969150794, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.38834677969150794}}
+:::MLLOG {"namespace": "", "time_ms": 1728521004546, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728521057409, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728521057409, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.40273311821800833, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.40273311821800833}}
+:::MLLOG {"namespace": "", "time_ms": 1728522209891, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728522262867, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728522262867, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.44312383646584586, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.44312383646584586}}
+:::MLLOG {"namespace": "", "time_ms": 1728523414970, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728523467811, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728523467812, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5062930049407103, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.5062930049407103}}
+:::MLLOG {"namespace": "", "time_ms": 1728524621021, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728524672880, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728524672880, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5888718058230566, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.5888718058230566}}
+:::MLLOG {"namespace": "", "time_ms": 1728525825012, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728525878025, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728525878025, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6833749091689574, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.6833749091689574}}
+:::MLLOG {"namespace": "", "time_ms": 1728527029872, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728527083655, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728527083655, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7040974767750154, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.7040974767750154}}
+:::MLLOG {"namespace": "", "time_ms": 1728528235285, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728528288182, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728528288183, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7093540651038799, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7093540651038799}}
+:::MLLOG {"namespace": "", "time_ms": 1728529439469, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728529492117, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728529492117, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7120359153824791, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7120359153824791}}
+:::MLLOG {"namespace": "", "time_ms": 1728530643295, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728530695998, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728530695999, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7137061499448043, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7137061499448043}}
+:::MLLOG {"namespace": "", "time_ms": 1728531854130, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728531905972, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728531905973, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7148870925168185, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7148870925168185}}
+:::MLLOG {"namespace": "", "time_ms": 1728533058059, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728533110923, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728533110923, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7155579556657943, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.7155579556657943}}
+:::MLLOG {"namespace": "", "time_ms": 1728534262752, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728534314582, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728534314582, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.716432806969261, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.716432806969261}}
+:::MLLOG {"namespace": "", "time_ms": 1728535466733, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728535518647, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728535518647, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7171081434700685, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7171081434700685}}
+:::MLLOG {"namespace": "", "time_ms": 1728536674344, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728536727210, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728536727211, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7175339546901564, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.7175339546901564}}
+:::MLLOG {"namespace": "", "time_ms": 1728537884762, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728537937482, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728537937482, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.718394822250531, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.718394822250531}}
+:::MLLOG {"namespace": "", "time_ms": 1728539088991, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2549184, "step_num": 38624}}
+:::MLLOG {"namespace": "", "time_ms": 1728539141647, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2549184, "step_num": 38624, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728539141648, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7188054349298025, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2549184, "masked_lm_accuracy": 0.7188054349298025}}
+:::MLLOG {"namespace": "", "time_ms": 1728540309614, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2699136, "step_num": 40896}}
+:::MLLOG {"namespace": "", "time_ms": 1728540362635, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2699136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2699136, "step_num": 40896, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728540362635, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7188799817474859, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2699136, "masked_lm_accuracy": 0.7188799817474859}}
+:::MLLOG {"namespace": "", "time_ms": 1728541513809, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2849088, "step_num": 43168}}
+:::MLLOG {"namespace": "", "time_ms": 1728541567687, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2849088, "step_num": 43168, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728541567687, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7195802817962523, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2849088, "masked_lm_accuracy": 0.7195802817962523}}
+:::MLLOG {"namespace": "", "time_ms": 1728542725822, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2999040, "step_num": 45440}}
+:::MLLOG {"namespace": "", "time_ms": 1728542778626, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2999040, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2999040, "step_num": 45440, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728542778627, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.719959929272595, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2999040, "masked_lm_accuracy": 0.719959929272595}}
+:::MLLOG {"namespace": "", "time_ms": 1728543937336, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3148992, "step_num": 47712}}
+:::MLLOG {"namespace": "", "time_ms": 1728543990349, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3148992, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3148992, "step_num": 47712, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728543990349, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7200285581392518, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3148992, "masked_lm_accuracy": 0.7200285581392518}}
+:::MLLOG {"namespace": "", "time_ms": 1728543990349, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 3148992, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 3148992}}
+:::MLLOG {"namespace": "", "time_ms": 1728543990349, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728543990349, "event_type": "POINT_IN_TIME", "key": "seed", "value": 21254, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,99 @@
+:::MLLOG {"namespace": "", "time_ms": 1728544003831, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728544003844, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_red", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728544003844, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728544003844, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728544003845, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728544004114, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728544004115, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728545567231, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728545577478, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728545591839, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728545591839, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728545591839, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728545591839, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728545591840, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728545591840, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728545591840, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728545591840, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728545591840, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728545591840, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728545591840, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728545591840, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728545591841, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728545591841, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728545591841, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728545591841, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728545591841, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728545644357, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728546841240, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728546898744, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728546898744, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.38783227444481694, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.38783227444481694}}
+:::MLLOG {"namespace": "", "time_ms": 1728548039569, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728548092787, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728548092788, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4103626927252508, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.4103626927252508}}
+:::MLLOG {"namespace": "", "time_ms": 1728549229970, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728549283151, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728549283152, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.45124479976326815, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.45124479976326815}}
+:::MLLOG {"namespace": "", "time_ms": 1728550420161, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728550473622, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728550473623, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5112407873616508, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.5112407873616508}}
+:::MLLOG {"namespace": "", "time_ms": 1728551611312, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728551664679, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728551664679, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6004241028086611, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.6004241028086611}}
+:::MLLOG {"namespace": "", "time_ms": 1728552801410, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728552853292, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728552853292, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6789556939443143, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.6789556939443143}}
+:::MLLOG {"namespace": "", "time_ms": 1728553989456, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728554042458, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728554042458, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7036405706305524, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.7036405706305524}}
+:::MLLOG {"namespace": "", "time_ms": 1728555178691, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728555231909, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728555231909, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7080815071178612, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7080815071178612}}
+:::MLLOG {"namespace": "", "time_ms": 1728556368078, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728556420173, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728556420173, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7108210982072117, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7108210982072117}}
+:::MLLOG {"namespace": "", "time_ms": 1728557556063, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728557608016, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728557608016, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7125928249628013, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7125928249628013}}
+:::MLLOG {"namespace": "", "time_ms": 1728558744467, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728558796348, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728558796348, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7135176017174266, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7135176017174266}}
+:::MLLOG {"namespace": "", "time_ms": 1728559932625, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728559984418, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728559984418, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7145863458576786, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.7145863458576786}}
+:::MLLOG {"namespace": "", "time_ms": 1728561120215, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728561173235, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728561173236, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7155747616727265, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.7155747616727265}}
+:::MLLOG {"namespace": "", "time_ms": 1728562309022, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728562360931, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728562360931, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7163686522768155, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7163686522768155}}
+:::MLLOG {"namespace": "", "time_ms": 1728563501949, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728563554781, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728563554781, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.716544431773359, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.716544431773359}}
+:::MLLOG {"namespace": "", "time_ms": 1728564690440, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728564743201, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728564743201, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7175336914214104, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.7175336914214104}}
+:::MLLOG {"namespace": "", "time_ms": 1728565883246, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2549184, "step_num": 38624}}
+:::MLLOG {"namespace": "", "time_ms": 1728565936169, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2549184, "step_num": 38624, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728565936169, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7186442701202992, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2549184, "masked_lm_accuracy": 0.7186442701202992}}
+:::MLLOG {"namespace": "", "time_ms": 1728567078310, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2699136, "step_num": 40896}}
+:::MLLOG {"namespace": "", "time_ms": 1728567130373, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2699136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2699136, "step_num": 40896, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728567130374, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7184390328784295, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2699136, "masked_lm_accuracy": 0.7184390328784295}}
+:::MLLOG {"namespace": "", "time_ms": 1728568270051, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2849088, "step_num": 43168}}
+:::MLLOG {"namespace": "", "time_ms": 1728568322854, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2849088, "step_num": 43168, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728568322854, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7190133258834455, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2849088, "masked_lm_accuracy": 0.7190133258834455}}
+:::MLLOG {"namespace": "", "time_ms": 1728569476727, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2999040, "step_num": 45440}}
+:::MLLOG {"namespace": "", "time_ms": 1728569528906, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2999040, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2999040, "step_num": 45440, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728569528906, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7193232258423117, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2999040, "masked_lm_accuracy": 0.7193232258423117}}
+:::MLLOG {"namespace": "", "time_ms": 1728570672300, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3148992, "step_num": 47712}}
+:::MLLOG {"namespace": "", "time_ms": 1728570725375, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3148992, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3148992, "step_num": 47712, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728570725375, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7193006322136452, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3148992, "masked_lm_accuracy": 0.7193006322136452}}
+:::MLLOG {"namespace": "", "time_ms": 1728571862039, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3298944, "step_num": 49984}}
+:::MLLOG {"namespace": "", "time_ms": 1728571914059, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3298944, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3298944, "step_num": 49984, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728571914060, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.71975367935961, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3298944, "masked_lm_accuracy": 0.71975367935961}}
+:::MLLOG {"namespace": "", "time_ms": 1728573050479, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3448896, "step_num": 52256}}
+:::MLLOG {"namespace": "", "time_ms": 1728573103486, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3448896, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3448896, "step_num": 52256, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728573103487, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7200944615325745, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3448896, "masked_lm_accuracy": 0.7200944615325745}}
+:::MLLOG {"namespace": "", "time_ms": 1728573103487, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 3448896, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 3448896}}
+:::MLLOG {"namespace": "", "time_ms": 1728573103487, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728573103487, "event_type": "POINT_IN_TIME", "key": "seed", "value": 31023, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,81 @@
+:::MLLOG {"namespace": "", "time_ms": 1728573115586, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728573115599, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_red", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728573115600, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728573115600, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728573115600, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728573115806, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728573115807, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728574701526, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728574711971, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728574726511, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728574726512, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728574726512, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728574726512, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728574726512, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728574726512, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728574726512, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728574726512, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728574726513, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728574726513, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728574726513, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728574726513, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728574726513, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728574726513, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728574726513, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728574726513, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728574726514, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728574767832, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728575974065, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728576033823, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728576033823, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.3871400942148816, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.3871400942148816}}
+:::MLLOG {"namespace": "", "time_ms": 1728577185498, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728577237616, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728577237616, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.39147048244665106, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.39147048244665106}}
+:::MLLOG {"namespace": "", "time_ms": 1728578385099, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728578438339, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728578438339, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.45783117753199354, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.45783117753199354}}
+:::MLLOG {"namespace": "", "time_ms": 1728579586799, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728579639018, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728579639018, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5536425763000323, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.5536425763000323}}
+:::MLLOG {"namespace": "", "time_ms": 1728580787082, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728580840252, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728580840252, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6808095073800067, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.6808095073800067}}
+:::MLLOG {"namespace": "", "time_ms": 1728581987844, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728582039916, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728582039917, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.70571623852529, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.70571623852529}}
+:::MLLOG {"namespace": "", "time_ms": 1728583186452, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728583239391, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728583239392, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7116154495584228, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.7116154495584228}}
+:::MLLOG {"namespace": "", "time_ms": 1728584385675, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728584438607, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728584438607, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7138893727993064, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7138893727993064}}
+:::MLLOG {"namespace": "", "time_ms": 1728585586366, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728585638448, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728585638449, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7147087185198344, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7147087185198344}}
+:::MLLOG {"namespace": "", "time_ms": 1728586785128, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728586837097, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728586837097, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7161258058628066, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7161258058628066}}
+:::MLLOG {"namespace": "", "time_ms": 1728587983727, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728588035691, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728588035692, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7175371565358254, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7175371565358254}}
+:::MLLOG {"namespace": "", "time_ms": 1728589188060, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728589241204, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728589241204, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.71796901310427, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.71796901310427}}
+:::MLLOG {"namespace": "", "time_ms": 1728590398050, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728590450110, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728590450111, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7184663905522461, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.7184663905522461}}
+:::MLLOG {"namespace": "", "time_ms": 1728591601749, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728591654909, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728591654909, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7191987109527519, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7191987109527519}}
+:::MLLOG {"namespace": "", "time_ms": 1728592801533, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728592853668, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728592853668, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7193377415935079, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.7193377415935079}}
+:::MLLOG {"namespace": "", "time_ms": 1728594003086, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728594056207, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728594056207, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7198944457577029, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.7198944457577029}}
+:::MLLOG {"namespace": "", "time_ms": 1728595208245, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2549184, "step_num": 38624}}
+:::MLLOG {"namespace": "", "time_ms": 1728595261648, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2549184, "step_num": 38624, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728595261648, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7206625282657168, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2549184, "masked_lm_accuracy": 0.7206625282657168}}
+:::MLLOG {"namespace": "", "time_ms": 1728595261648, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 2549184}}
+:::MLLOG {"namespace": "", "time_ms": 1728595261648, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728595261648, "event_type": "POINT_IN_TIME", "key": "seed", "value": 6117, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,90 @@
+:::MLLOG {"namespace": "", "time_ms": 1728595273658, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728595273672, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_red", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728595273672, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728595273672, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728595273672, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728595273864, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728595273865, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728596851159, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728596861539, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728596876123, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728596876124, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728596876124, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728596876124, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728596876124, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728596876124, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728596876124, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728596876125, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728596876125, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728596876125, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728596876125, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728596876125, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728596876125, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728596876125, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728596876126, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728596876126, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728596876126, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728596921065, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728598131675, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728598191093, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728598191093, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.38629304813852405, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.38629304813852405}}
+:::MLLOG {"namespace": "", "time_ms": 1728599345629, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728599399811, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728599399812, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4057833703070539, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.4057833703070539}}
+:::MLLOG {"namespace": "", "time_ms": 1728600550664, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728600603826, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728600603826, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4394682460404758, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.4394682460404758}}
+:::MLLOG {"namespace": "", "time_ms": 1728601755405, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728601807545, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728601807545, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5091635388711576, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.5091635388711576}}
+:::MLLOG {"namespace": "", "time_ms": 1728602959507, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728603011549, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728603011550, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6130874210764136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.6130874210764136}}
+:::MLLOG {"namespace": "", "time_ms": 1728604163051, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728604215106, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728604215107, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6886828500994252, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.6886828500994252}}
+:::MLLOG {"namespace": "", "time_ms": 1728605365869, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728605418812, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728605418812, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7043221683007339, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.7043221683007339}}
+:::MLLOG {"namespace": "", "time_ms": 1728606570227, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728606622252, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728606622252, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7098102495208356, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7098102495208356}}
+:::MLLOG {"namespace": "", "time_ms": 1728607773165, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728607826339, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728607826339, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7117401148361865, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7117401148361865}}
+:::MLLOG {"namespace": "", "time_ms": 1728608976638, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728609028698, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728609028698, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7135091588821822, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7135091588821822}}
+:::MLLOG {"namespace": "", "time_ms": 1728610184971, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728610238021, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728610238021, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.714804682570013, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.714804682570013}}
+:::MLLOG {"namespace": "", "time_ms": 1728611388386, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728611441326, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728611441326, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7155881319921319, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.7155881319921319}}
+:::MLLOG {"namespace": "", "time_ms": 1728612596531, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728612648484, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728612648485, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7166956780410199, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.7166956780410199}}
+:::MLLOG {"namespace": "", "time_ms": 1728613799458, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728613851376, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728613851376, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7175573623673818, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7175573623673818}}
+:::MLLOG {"namespace": "", "time_ms": 1728615017853, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728615070888, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728615070888, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7177324852295528, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.7177324852295528}}
+:::MLLOG {"namespace": "", "time_ms": 1728616224796, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728616276794, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728616276794, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7187149448529216, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.7187149448529216}}
+:::MLLOG {"namespace": "", "time_ms": 1728617427166, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2549184, "step_num": 38624}}
+:::MLLOG {"namespace": "", "time_ms": 1728617480097, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2549184, "step_num": 38624, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728617480098, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7189440351084598, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2549184, "masked_lm_accuracy": 0.7189440351084598}}
+:::MLLOG {"namespace": "", "time_ms": 1728618630518, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2699136, "step_num": 40896}}
+:::MLLOG {"namespace": "", "time_ms": 1728618683523, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2699136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2699136, "step_num": 40896, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728618683524, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7189651959015355, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2699136, "masked_lm_accuracy": 0.7189651959015355}}
+:::MLLOG {"namespace": "", "time_ms": 1728619834438, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2849088, "step_num": 43168}}
+:::MLLOG {"namespace": "", "time_ms": 1728619886484, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2849088, "step_num": 43168, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728619886484, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7197905807489396, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2849088, "masked_lm_accuracy": 0.7197905807489396}}
+:::MLLOG {"namespace": "", "time_ms": 1728621047498, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2999040, "step_num": 45440}}
+:::MLLOG {"namespace": "", "time_ms": 1728621100440, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2999040, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2999040, "step_num": 45440, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728621100440, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7202326569216796, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2999040, "masked_lm_accuracy": 0.7202326569216796}}
+:::MLLOG {"namespace": "", "time_ms": 1728621100441, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 2999040, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 2999040}}
+:::MLLOG {"namespace": "", "time_ms": 1728621100441, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728621100441, "event_type": "POINT_IN_TIME", "key": "seed", "value": 18962, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,84 @@
+:::MLLOG {"namespace": "", "time_ms": 1728621113036, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728621113050, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_red", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728621113050, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728621113050, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728621113050, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728621113364, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728621113364, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728622708323, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728622718838, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728622734010, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728622734010, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728622734010, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728622734011, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728622734011, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728622734011, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728622734011, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728622734011, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728622734011, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728622734012, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728622734012, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728622734012, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728622734012, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728622734012, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728622734012, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728622734012, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728622734013, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728622779756, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728623989242, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728624047931, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728624047931, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.38761837577562386, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.38761837577562386}}
+:::MLLOG {"namespace": "", "time_ms": 1728625199877, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728625251862, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728625251862, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.3991255365093096, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.3991255365093096}}
+:::MLLOG {"namespace": "", "time_ms": 1728626401253, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728626454238, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728626454239, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4629036359752662, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.4629036359752662}}
+:::MLLOG {"namespace": "", "time_ms": 1728627603380, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728627655201, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728627655201, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5288661295558138, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.5288661295558138}}
+:::MLLOG {"namespace": "", "time_ms": 1728628804682, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728628857572, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728628857573, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6336066501184932, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.6336066501184932}}
+:::MLLOG {"namespace": "", "time_ms": 1728630005680, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728630057532, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728630057532, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.698150883303049, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.698150883303049}}
+:::MLLOG {"namespace": "", "time_ms": 1728631206492, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728631258444, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728631258444, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7087594178456637, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.7087594178456637}}
+:::MLLOG {"namespace": "", "time_ms": 1728632406126, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728632458945, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728632458946, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.712074551420721, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.712074551420721}}
+:::MLLOG {"namespace": "", "time_ms": 1728633607305, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728633660325, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728633660325, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7133723366167564, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7133723366167564}}
+:::MLLOG {"namespace": "", "time_ms": 1728634807043, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728634859650, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728634859650, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7153185129380183, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7153185129380183}}
+:::MLLOG {"namespace": "", "time_ms": 1728636007670, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728636060454, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728636060454, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7153358636105497, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7153358636105497}}
+:::MLLOG {"namespace": "", "time_ms": 1728637212672, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728637266430, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728637266431, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7166292644147753, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.7166292644147753}}
+:::MLLOG {"namespace": "", "time_ms": 1728638413489, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728638466302, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728638466303, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.717635804439778, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.717635804439778}}
+:::MLLOG {"namespace": "", "time_ms": 1728639621382, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728639673232, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728639673232, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7184983831576123, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7184983831576123}}
+:::MLLOG {"namespace": "", "time_ms": 1728640830900, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728640882936, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728640882936, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7185333441243461, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.7185333441243461}}
+:::MLLOG {"namespace": "", "time_ms": 1728642038106, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728642090127, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728642090127, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.71883562362759, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.71883562362759}}
+:::MLLOG {"namespace": "", "time_ms": 1728643238601, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2549184, "step_num": 38624}}
+:::MLLOG {"namespace": "", "time_ms": 1728643291575, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2549184, "step_num": 38624, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728643291575, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7195813601862262, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2549184, "masked_lm_accuracy": 0.7195813601862262}}
+:::MLLOG {"namespace": "", "time_ms": 1728644439777, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2699136, "step_num": 40896}}
+:::MLLOG {"namespace": "", "time_ms": 1728644491633, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2699136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2699136, "step_num": 40896, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728644491633, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7200058278167899, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2699136, "masked_lm_accuracy": 0.7200058278167899}}
+:::MLLOG {"namespace": "", "time_ms": 1728644491633, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 2699136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 2699136}}
+:::MLLOG {"namespace": "", "time_ms": 1728644491633, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728644491633, "event_type": "POINT_IN_TIME", "key": "seed", "value": 7451, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,90 @@
+:::MLLOG {"namespace": "", "time_ms": 1728524927811, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728524927825, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_red", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728524927825, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728524927825, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728524927825, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728524927826, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728524927826, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728526542156, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728526553341, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728526570472, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728526570473, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728526570473, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728526570473, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728526570473, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728526570473, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728526570473, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728526570474, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728526570474, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728526570474, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728526570474, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728526570474, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728526570474, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728526570474, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728526570474, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728526570475, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728526570475, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728526620720, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728527836136, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728527898691, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728527898691, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.3863042194326981, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.3863042194326981}}
+:::MLLOG {"namespace": "", "time_ms": 1728529058617, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728529115079, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728529115080, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.40961273735915393, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.40961273735915393}}
+:::MLLOG {"namespace": "", "time_ms": 1728530272495, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728530328049, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728530328049, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.43488811103970115, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.43488811103970115}}
+:::MLLOG {"namespace": "", "time_ms": 1728531485648, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728531541201, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728531541201, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4996365338081218, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.4996365338081218}}
+:::MLLOG {"namespace": "", "time_ms": 1728532697882, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728532753299, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728532753299, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5947849176998402, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.5947849176998402}}
+:::MLLOG {"namespace": "", "time_ms": 1728533910552, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728533966102, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728533966102, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.690821249612306, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.690821249612306}}
+:::MLLOG {"namespace": "", "time_ms": 1728535121207, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728535177526, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728535177526, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7054878836201564, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.7054878836201564}}
+:::MLLOG {"namespace": "", "time_ms": 1728536332776, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728536389219, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728536389220, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7097416914145819, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7097416914145819}}
+:::MLLOG {"namespace": "", "time_ms": 1728537544044, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728537599323, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728537599324, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7116119135763378, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7116119135763378}}
+:::MLLOG {"namespace": "", "time_ms": 1728538754453, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728538809839, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728538809839, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7137712529124653, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7137712529124653}}
+:::MLLOG {"namespace": "", "time_ms": 1728539966085, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728540021340, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728540021340, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7149044781941172, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7149044781941172}}
+:::MLLOG {"namespace": "", "time_ms": 1728541177404, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728541233606, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728541233606, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7153092085349753, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.7153092085349753}}
+:::MLLOG {"namespace": "", "time_ms": 1728542397120, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728542452484, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728542452485, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.716667587340915, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.716667587340915}}
+:::MLLOG {"namespace": "", "time_ms": 1728543613320, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728543669699, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728543669699, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7173634540126506, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7173634540126506}}
+:::MLLOG {"namespace": "", "time_ms": 1728544826341, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728544881826, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728544881826, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7179233225148527, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.7179233225148527}}
+:::MLLOG {"namespace": "", "time_ms": 1728546038641, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728546093965, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728546093965, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7184311815844229, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.7184311815844229}}
+:::MLLOG {"namespace": "", "time_ms": 1728547260318, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2549184, "step_num": 38624}}
+:::MLLOG {"namespace": "", "time_ms": 1728547316564, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2549184, "step_num": 38624, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728547316565, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7191096605050328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2549184, "masked_lm_accuracy": 0.7191096605050328}}
+:::MLLOG {"namespace": "", "time_ms": 1728548476862, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2699136, "step_num": 40896}}
+:::MLLOG {"namespace": "", "time_ms": 1728548532277, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2699136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2699136, "step_num": 40896, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728548532277, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7192467427854418, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2699136, "masked_lm_accuracy": 0.7192467427854418}}
+:::MLLOG {"namespace": "", "time_ms": 1728549694358, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2849088, "step_num": 43168}}
+:::MLLOG {"namespace": "", "time_ms": 1728549750718, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2849088, "step_num": 43168, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728549750718, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7199938913341333, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2849088, "masked_lm_accuracy": 0.7199938913341333}}
+:::MLLOG {"namespace": "", "time_ms": 1728550912201, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2999040, "step_num": 45440}}
+:::MLLOG {"namespace": "", "time_ms": 1728550968747, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2999040, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2999040, "step_num": 45440, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728550968748, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7203293324374981, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2999040, "masked_lm_accuracy": 0.7203293324374981}}
+:::MLLOG {"namespace": "", "time_ms": 1728550968748, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 2999040, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 2999040}}
+:::MLLOG {"namespace": "", "time_ms": 1728550968748, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728550968748, "event_type": "POINT_IN_TIME", "key": "seed", "value": 31643, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,96 @@
+:::MLLOG {"namespace": "", "time_ms": 1728550982004, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728550982018, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_red", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728550982018, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728550982018, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728550982018, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728550982200, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728550982201, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728552594185, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728552604676, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728552619008, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728552619008, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728552619008, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728552619009, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728552619009, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728552619009, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728552619009, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728552619009, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728552619009, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728552619009, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728552619010, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728552619010, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728552619010, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728552619011, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728552619011, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728552619011, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728552619011, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728552666986, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728553875969, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728553939951, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728553939951, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.38616175677830683, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.38616175677830683}}
+:::MLLOG {"namespace": "", "time_ms": 1728555092364, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728555150488, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728555150489, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.41162710727584095, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.41162710727584095}}
+:::MLLOG {"namespace": "", "time_ms": 1728556296541, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728556355574, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728556355575, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4350904176781545, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.4350904176781545}}
+:::MLLOG {"namespace": "", "time_ms": 1728557502466, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728557561747, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728557561747, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.496310165579904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.496310165579904}}
+:::MLLOG {"namespace": "", "time_ms": 1728558708887, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728558766886, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728558766887, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5802412135699729, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.5802412135699729}}
+:::MLLOG {"namespace": "", "time_ms": 1728559914092, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728559973260, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728559973260, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6652080704154694, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.6652080704154694}}
+:::MLLOG {"namespace": "", "time_ms": 1728561120077, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728561179197, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728561179198, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6984256211339748, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.6984256211339748}}
+:::MLLOG {"namespace": "", "time_ms": 1728562325103, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728562383110, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728562383111, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7061839626088569, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7061839626088569}}
+:::MLLOG {"namespace": "", "time_ms": 1728563529512, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728563588342, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728563588343, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.709852535744663, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.709852535744663}}
+:::MLLOG {"namespace": "", "time_ms": 1728564732742, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728564791651, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728564791651, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7124425528931918, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7124425528931918}}
+:::MLLOG {"namespace": "", "time_ms": 1728565936303, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728565995377, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728565995377, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7139212561175718, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7139212561175718}}
+:::MLLOG {"namespace": "", "time_ms": 1728567148853, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728567207007, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728567207007, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7147917833311085, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.7147917833311085}}
+:::MLLOG {"namespace": "", "time_ms": 1728568352291, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728568410261, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728568410261, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7158374547529306, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.7158374547529306}}
+:::MLLOG {"namespace": "", "time_ms": 1728569560993, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728569619018, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728569619018, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7164395152271045, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7164395152271045}}
+:::MLLOG {"namespace": "", "time_ms": 1728570770470, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728570828476, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728570828476, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7171282240496328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.7171282240496328}}
+:::MLLOG {"namespace": "", "time_ms": 1728571971989, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728572031038, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728572031039, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7176245095920047, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.7176245095920047}}
+:::MLLOG {"namespace": "", "time_ms": 1728573181122, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2549184, "step_num": 38624}}
+:::MLLOG {"namespace": "", "time_ms": 1728573239069, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2549184, "step_num": 38624, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728573239069, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7185501086189852, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2549184, "masked_lm_accuracy": 0.7185501086189852}}
+:::MLLOG {"namespace": "", "time_ms": 1728574384118, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2699136, "step_num": 40896}}
+:::MLLOG {"namespace": "", "time_ms": 1728574443334, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2699136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2699136, "step_num": 40896, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728574443334, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.718467390029532, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2699136, "masked_lm_accuracy": 0.718467390029532}}
+:::MLLOG {"namespace": "", "time_ms": 1728575608940, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2849088, "step_num": 43168}}
+:::MLLOG {"namespace": "", "time_ms": 1728575667042, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2849088, "step_num": 43168, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728575667042, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7193740987105504, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2849088, "masked_lm_accuracy": 0.7193740987105504}}
+:::MLLOG {"namespace": "", "time_ms": 1728576812654, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2999040, "step_num": 45440}}
+:::MLLOG {"namespace": "", "time_ms": 1728576871686, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2999040, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2999040, "step_num": 45440, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728576871686, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.719547293968521, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2999040, "masked_lm_accuracy": 0.719547293968521}}
+:::MLLOG {"namespace": "", "time_ms": 1728578016561, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3148992, "step_num": 47712}}
+:::MLLOG {"namespace": "", "time_ms": 1728578075504, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3148992, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3148992, "step_num": 47712, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728578075504, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7198685481652716, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3148992, "masked_lm_accuracy": 0.7198685481652716}}
+:::MLLOG {"namespace": "", "time_ms": 1728579221384, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 3298944, "step_num": 49984}}
+:::MLLOG {"namespace": "", "time_ms": 1728579280495, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 3298944, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 3298944, "step_num": 49984, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728579280495, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7202844063393284, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 3298944, "masked_lm_accuracy": 0.7202844063393284}}
+:::MLLOG {"namespace": "", "time_ms": 1728579280496, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 3298944, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 3298944}}
+:::MLLOG {"namespace": "", "time_ms": 1728579280496, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728579280496, "event_type": "POINT_IN_TIME", "key": "seed", "value": 26715, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,87 @@
+:::MLLOG {"namespace": "", "time_ms": 1728579292347, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728579292360, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_red", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728579292360, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728579292360, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728579292360, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728579292510, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728579292511, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728580879014, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728580889187, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728580903804, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728580903804, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728580903805, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728580903805, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728580903805, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728580903805, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728580903805, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728580903805, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728580903805, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728580903806, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728580903806, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728580903806, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728580903806, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728580903806, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728580903806, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728580903806, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728580903806, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728580951866, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728582174949, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728582235954, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728582235955, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.38839849547204247, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.38839849547204247}}
+:::MLLOG {"namespace": "", "time_ms": 1728583404398, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728583457118, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728583457118, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4043779520660943, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.4043779520660943}}
+:::MLLOG {"namespace": "", "time_ms": 1728584623300, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728584677358, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728584677358, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4277538998833515, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.4277538998833515}}
+:::MLLOG {"namespace": "", "time_ms": 1728585841710, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728585894585, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728585894586, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4895906483404781, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.4895906483404781}}
+:::MLLOG {"namespace": "", "time_ms": 1728587058762, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728587112509, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728587112509, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5868821954505488, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.5868821954505488}}
+:::MLLOG {"namespace": "", "time_ms": 1728588276137, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728588329834, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728588329834, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6855756070131875, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.6855756070131875}}
+:::MLLOG {"namespace": "", "time_ms": 1728589494361, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728589548276, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728589548276, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7039273525328428, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.7039273525328428}}
+:::MLLOG {"namespace": "", "time_ms": 1728590710884, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728590765567, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728590765567, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7093820230075536, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7093820230075536}}
+:::MLLOG {"namespace": "", "time_ms": 1728591929060, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728591981842, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728591981842, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7118489585669369, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7118489585669369}}
+:::MLLOG {"namespace": "", "time_ms": 1728593146668, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728593199377, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728593199377, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7138080239009914, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7138080239009914}}
+:::MLLOG {"namespace": "", "time_ms": 1728594362906, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728594416691, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728594416692, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7149828011287352, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7149828011287352}}
+:::MLLOG {"namespace": "", "time_ms": 1728595585602, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728595639450, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728595639450, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7161310530023511, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.7161310530023511}}
+:::MLLOG {"namespace": "", "time_ms": 1728596808968, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728596861749, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728596861749, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7167342547606621, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.7167342547606621}}
+:::MLLOG {"namespace": "", "time_ms": 1728598025777, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728598079566, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728598079567, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7175075857669347, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7175075857669347}}
+:::MLLOG {"namespace": "", "time_ms": 1728599242962, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728599297649, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728599297649, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7177893112502416, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.7177893112502416}}
+:::MLLOG {"namespace": "", "time_ms": 1728600461272, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728600514019, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728600514019, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7186980957604484, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.7186980957604484}}
+:::MLLOG {"namespace": "", "time_ms": 1728601688118, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2549184, "step_num": 38624}}
+:::MLLOG {"namespace": "", "time_ms": 1728601741703, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2549184, "step_num": 38624, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728601741703, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7194635579882086, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2549184, "masked_lm_accuracy": 0.7194635579882086}}
+:::MLLOG {"namespace": "", "time_ms": 1728602912396, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2699136, "step_num": 40896}}
+:::MLLOG {"namespace": "", "time_ms": 1728602965140, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2699136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2699136, "step_num": 40896, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728602965140, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7196329377980453, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2699136, "masked_lm_accuracy": 0.7196329377980453}}
+:::MLLOG {"namespace": "", "time_ms": 1728604140305, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2849088, "step_num": 43168}}
+:::MLLOG {"namespace": "", "time_ms": 1728604193177, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2849088, "step_num": 43168, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728604193177, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7201142786646146, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2849088, "masked_lm_accuracy": 0.7201142786646146}}
+:::MLLOG {"namespace": "", "time_ms": 1728604193178, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 2849088}}
+:::MLLOG {"namespace": "", "time_ms": 1728604193178, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728604193178, "event_type": "POINT_IN_TIME", "key": "seed", "value": 24659, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,90 @@
+:::MLLOG {"namespace": "", "time_ms": 1728604206238, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728604206251, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_red", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728604206251, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728604206251, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728604206251, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728604206399, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728604206400, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728605831206, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728605841697, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728605856169, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728605856169, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728605856170, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728605856170, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728605856170, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728605856170, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728605856170, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728605856170, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728605856170, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728605856171, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728605856171, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728605856171, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728605856171, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728605856171, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728605856171, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728605856171, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728605856171, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728605907083, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728607116687, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728607176173, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728607176173, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.38733550816792245, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.38733550816792245}}
+:::MLLOG {"namespace": "", "time_ms": 1728608331645, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728608385445, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728608385445, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.4099236127138853, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.4099236127138853}}
+:::MLLOG {"namespace": "", "time_ms": 1728609536865, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728609589476, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728609589476, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.46445941542463526, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.46445941542463526}}
+:::MLLOG {"namespace": "", "time_ms": 1728610742733, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728610796400, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728610796401, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5356849022231992, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.5356849022231992}}
+:::MLLOG {"namespace": "", "time_ms": 1728611950463, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728612003989, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728612003989, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6243599510936588, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.6243599510936588}}
+:::MLLOG {"namespace": "", "time_ms": 1728613157626, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728613211058, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728613211058, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6915194793740074, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.6915194793740074}}
+:::MLLOG {"namespace": "", "time_ms": 1728614363023, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728614416397, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728614416397, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.70565997331387, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.70565997331387}}
+:::MLLOG {"namespace": "", "time_ms": 1728615568690, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728615622127, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728615622128, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7094783287385873, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7094783287385873}}
+:::MLLOG {"namespace": "", "time_ms": 1728616773489, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728616825919, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728616825919, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7119873171805191, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7119873171805191}}
+:::MLLOG {"namespace": "", "time_ms": 1728617978563, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728618032221, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728618032222, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7142333673658525, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7142333673658525}}
+:::MLLOG {"namespace": "", "time_ms": 1728619183815, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728619236305, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728619236305, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7149768588638763, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7149768588638763}}
+:::MLLOG {"namespace": "", "time_ms": 1728620386747, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728620439998, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728620439998, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7156523679928931, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.7156523679928931}}
+:::MLLOG {"namespace": "", "time_ms": 1728621597193, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728621650622, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728621650622, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7164913090174971, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.7164913090174971}}
+:::MLLOG {"namespace": "", "time_ms": 1728622809144, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728622861570, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728622861570, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7172874884828523, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7172874884828523}}
+:::MLLOG {"namespace": "", "time_ms": 1728624015271, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728624067817, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728624067818, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7180004538213985, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.7180004538213985}}
+:::MLLOG {"namespace": "", "time_ms": 1728625221988, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728625275277, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728625275277, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7182580689577264, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.7182580689577264}}
+:::MLLOG {"namespace": "", "time_ms": 1728626428374, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2549184, "step_num": 38624}}
+:::MLLOG {"namespace": "", "time_ms": 1728626481696, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2549184, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2549184, "step_num": 38624, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728626481696, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7189949285361891, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2549184, "masked_lm_accuracy": 0.7189949285361891}}
+:::MLLOG {"namespace": "", "time_ms": 1728627644056, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2699136, "step_num": 40896}}
+:::MLLOG {"namespace": "", "time_ms": 1728627696455, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2699136, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2699136, "step_num": 40896, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728627696456, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7192161206220823, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2699136, "masked_lm_accuracy": 0.7192161206220823}}
+:::MLLOG {"namespace": "", "time_ms": 1728628864498, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2849088, "step_num": 43168}}
+:::MLLOG {"namespace": "", "time_ms": 1728628917907, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2849088, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2849088, "step_num": 43168, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728628917908, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7198127043697744, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2849088, "masked_lm_accuracy": 0.7198127043697744}}
+:::MLLOG {"namespace": "", "time_ms": 1728630073687, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2999040, "step_num": 45440}}
+:::MLLOG {"namespace": "", "time_ms": 1728630126070, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2999040, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2999040, "step_num": 45440, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728630126070, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7204812415431342, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2999040, "masked_lm_accuracy": 0.7204812415431342}}
+:::MLLOG {"namespace": "", "time_ms": 1728630126070, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 2999040, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 2999040}}
+:::MLLOG {"namespace": "", "time_ms": 1728630126071, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728630126071, "event_type": "POINT_IN_TIME", "key": "seed", "value": 6018, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -0,0 +1,78 @@
+:::MLLOG {"namespace": "", "time_ms": 1728630138587, "event_type": "POINT_IN_TIME", "key": "submission_org", "value": "tinycorp", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 631}}
+:::MLLOG {"namespace": "", "time_ms": 1728630138600, "event_type": "POINT_IN_TIME", "key": "submission_platform", "value": "tinybox_red", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 632}}
+:::MLLOG {"namespace": "", "time_ms": 1728630138600, "event_type": "POINT_IN_TIME", "key": "submission_division", "value": "closed", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 633}}
+:::MLLOG {"namespace": "", "time_ms": 1728630138601, "event_type": "POINT_IN_TIME", "key": "submission_status", "value": "onprem", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 634}}
+:::MLLOG {"namespace": "", "time_ms": 1728630138601, "event_type": "POINT_IN_TIME", "key": "submission_benchmark", "value": "bert", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 636}}
+:::MLLOG {"namespace": "", "time_ms": 1728630138750, "event_type": "POINT_IN_TIME", "key": "cache_clear", "value": true, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 639}}
+:::MLLOG {"namespace": "", "time_ms": 1728630138750, "event_type": "INTERVAL_START", "key": "init_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 640}}
+:::MLLOG {"namespace": "", "time_ms": 1728631762258, "event_type": "POINT_IN_TIME", "key": "init_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 842}}
+:::MLLOG {"namespace": "", "time_ms": 1728631772243, "event_type": "INTERVAL_START", "key": "run_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 643}}
+:::MLLOG {"namespace": "", "time_ms": 1728631787279, "event_type": "POINT_IN_TIME", "key": "global_batch_size", "value": 66, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 711}}
+:::MLLOG {"namespace": "", "time_ms": 1728631787280, "event_type": "POINT_IN_TIME", "key": "max_sequence_length", "value": 512, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 712}}
+:::MLLOG {"namespace": "", "time_ms": 1728631787280, "event_type": "POINT_IN_TIME", "key": "max_predictions_per_seq", "value": 76, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 713}}
+:::MLLOG {"namespace": "", "time_ms": 1728631787280, "event_type": "POINT_IN_TIME", "key": "opt_name", "value": "LAMB", "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 715}}
+:::MLLOG {"namespace": "", "time_ms": 1728631787280, "event_type": "POINT_IN_TIME", "key": "opt_base_learning_rate", "value": 0.0001, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 716}}
+:::MLLOG {"namespace": "", "time_ms": 1728631787280, "event_type": "POINT_IN_TIME", "key": "opt_lamb_weight_decay_rate", "value": 0.01, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 717}}
+:::MLLOG {"namespace": "", "time_ms": 1728631787280, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_1", "value": 0.9, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 718}}
+:::MLLOG {"namespace": "", "time_ms": 1728631787281, "event_type": "POINT_IN_TIME", "key": "opt_lamb_beta_2", "value": 0.999, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 719}}
+:::MLLOG {"namespace": "", "time_ms": 1728631787281, "event_type": "POINT_IN_TIME", "key": "opt_lamb_learning_rate_decay_poly_power", "value": 1.0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 720}}
+:::MLLOG {"namespace": "", "time_ms": 1728631787281, "event_type": "POINT_IN_TIME", "key": "opt_lamb_epsilon", "value": 1e-06, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 721}}
+:::MLLOG {"namespace": "", "time_ms": 1728631787281, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 723}}
+:::MLLOG {"namespace": "", "time_ms": 1728631787281, "event_type": "POINT_IN_TIME", "key": "num_warmup_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 724}}
+:::MLLOG {"namespace": "", "time_ms": 1728631787281, "event_type": "POINT_IN_TIME", "key": "start_warmup_step", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 725}}
+:::MLLOG {"namespace": "", "time_ms": 1728631787281, "event_type": "POINT_IN_TIME", "key": "opt_learning_rate_training_steps", "value": 55000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 726}}
+:::MLLOG {"namespace": "", "time_ms": 1728631787282, "event_type": "POINT_IN_TIME", "key": "gradient_accumulation_steps", "value": 1, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 727}}
+:::MLLOG {"namespace": "", "time_ms": 1728631787282, "event_type": "POINT_IN_TIME", "key": "eval_samples", "value": 10002, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 728}}
+:::MLLOG {"namespace": "", "time_ms": 1728631787282, "event_type": "POINT_IN_TIME", "key": "train_samples", "value": 3630000, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 729}}
+:::MLLOG {"namespace": "", "time_ms": 1728631833698, "event_type": "INTERVAL_START", "key": "epoch_start", "value": 0, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 753, "epoch_num": 0}}
+:::MLLOG {"namespace": "", "time_ms": 1728633040032, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 149952, "step_num": 2272}}
+:::MLLOG {"namespace": "", "time_ms": 1728633099704, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 149952, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 149952, "step_num": 2272, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728633099705, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.38797098084452913, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 149952, "masked_lm_accuracy": 0.38797098084452913}}
+:::MLLOG {"namespace": "", "time_ms": 1728634252081, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 299904, "step_num": 4544}}
+:::MLLOG {"namespace": "", "time_ms": 1728634305117, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 299904, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 299904, "step_num": 4544, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728634305117, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.3995994811617263, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 299904, "masked_lm_accuracy": 0.3995994811617263}}
+:::MLLOG {"namespace": "", "time_ms": 1728635455137, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 449856, "step_num": 6816}}
+:::MLLOG {"namespace": "", "time_ms": 1728635508121, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 449856, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 449856, "step_num": 6816, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728635508121, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.5011176564375941, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 449856, "masked_lm_accuracy": 0.5011176564375941}}
+:::MLLOG {"namespace": "", "time_ms": 1728636656616, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 599808, "step_num": 9088}}
+:::MLLOG {"namespace": "", "time_ms": 1728636710601, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 599808, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 599808, "step_num": 9088, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728636710602, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.6374410459147146, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 599808, "masked_lm_accuracy": 0.6374410459147146}}
+:::MLLOG {"namespace": "", "time_ms": 1728637859253, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 749760, "step_num": 11360}}
+:::MLLOG {"namespace": "", "time_ms": 1728637913116, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 749760, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 749760, "step_num": 11360, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728637913117, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7006531824018688, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 749760, "masked_lm_accuracy": 0.7006531824018688}}
+:::MLLOG {"namespace": "", "time_ms": 1728639062400, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 899712, "step_num": 13632}}
+:::MLLOG {"namespace": "", "time_ms": 1728639116398, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 899712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 899712, "step_num": 13632, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728639116398, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7092244053358937, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 899712, "masked_lm_accuracy": 0.7092244053358937}}
+:::MLLOG {"namespace": "", "time_ms": 1728640264599, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1049664, "step_num": 15904}}
+:::MLLOG {"namespace": "", "time_ms": 1728640317583, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1049664, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1049664, "step_num": 15904, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728640317584, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7131378053546167, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1049664, "masked_lm_accuracy": 0.7131378053546167}}
+:::MLLOG {"namespace": "", "time_ms": 1728641465346, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1199616, "step_num": 18176}}
+:::MLLOG {"namespace": "", "time_ms": 1728641518225, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1199616, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1199616, "step_num": 18176, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728641518225, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7143681931080901, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1199616, "masked_lm_accuracy": 0.7143681931080901}}
+:::MLLOG {"namespace": "", "time_ms": 1728642666024, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1349568, "step_num": 20448}}
+:::MLLOG {"namespace": "", "time_ms": 1728642718905, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1349568, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1349568, "step_num": 20448, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728642718905, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7157575550305322, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1349568, "masked_lm_accuracy": 0.7157575550305322}}
+:::MLLOG {"namespace": "", "time_ms": 1728643864313, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1499520, "step_num": 22720}}
+:::MLLOG {"namespace": "", "time_ms": 1728643917070, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1499520, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1499520, "step_num": 22720, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728643917070, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7172654897564532, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1499520, "masked_lm_accuracy": 0.7172654897564532}}
+:::MLLOG {"namespace": "", "time_ms": 1728645063069, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1649472, "step_num": 24992}}
+:::MLLOG {"namespace": "", "time_ms": 1728645115906, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1649472, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1649472, "step_num": 24992, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728645115907, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7175817649094158, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1649472, "masked_lm_accuracy": 0.7175817649094158}}
+:::MLLOG {"namespace": "", "time_ms": 1728646261029, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1799424, "step_num": 27264}}
+:::MLLOG {"namespace": "", "time_ms": 1728646314945, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1799424, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1799424, "step_num": 27264, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728646314946, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7180203909851078, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1799424, "masked_lm_accuracy": 0.7180203909851078}}
+:::MLLOG {"namespace": "", "time_ms": 1728647468431, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 1949376, "step_num": 29536}}
+:::MLLOG {"namespace": "", "time_ms": 1728647522632, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 1949376, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 1949376, "step_num": 29536, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728647522633, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7191875515115712, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 1949376, "masked_lm_accuracy": 0.7191875515115712}}
+:::MLLOG {"namespace": "", "time_ms": 1728648686121, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2099328, "step_num": 31808}}
+:::MLLOG {"namespace": "", "time_ms": 1728648739069, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2099328, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2099328, "step_num": 31808, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728648739069, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7196067409309428, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2099328, "masked_lm_accuracy": 0.7196067409309428}}
+:::MLLOG {"namespace": "", "time_ms": 1728649885357, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2249280, "step_num": 34080}}
+:::MLLOG {"namespace": "", "time_ms": 1728649939244, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2249280, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2249280, "step_num": 34080, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728649939245, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7197489482454004, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2249280, "masked_lm_accuracy": 0.7197489482454004}}
+:::MLLOG {"namespace": "", "time_ms": 1728651085252, "event_type": "INTERVAL_START", "key": "eval_start", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 806, "epoch_num": 2399232, "step_num": 36352}}
+:::MLLOG {"namespace": "", "time_ms": 1728651139304, "event_type": "INTERVAL_END", "key": "eval_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 861, "epoch_count": 2399232, "step_num": 36352, "samples_count": 10002}}
+:::MLLOG {"namespace": "", "time_ms": 1728651139304, "event_type": "POINT_IN_TIME", "key": "eval_accuracy", "value": 0.7202936197442785, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 862, "epoch_num": 2399232, "masked_lm_accuracy": 0.7202936197442785}}
+:::MLLOG {"namespace": "", "time_ms": 1728651139305, "event_type": "POINT_IN_TIME", "key": "epoch_stop", "value": 2399232, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 880, "epoch_num": 2399232}}
+:::MLLOG {"namespace": "", "time_ms": 1728651139305, "event_type": "INTERVAL_END", "key": "run_stop", "value": null, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 881, "status": "success"}}
+:::MLLOG {"namespace": "", "time_ms": 1728651139305, "event_type": "POINT_IN_TIME", "key": "seed", "value": 32021, "metadata": {"file": "tinygrad/examples/mlperf/model_train.py", "lineno": 882}}
@@ -1,73 +0,0 @@
-# 1. Problem
-
-This problem uses BERT for NLP.
-
-## Requirements
-
-Install tinygrad and mlperf-logging from master.
-```
-git clone https://github.com/tinygrad/tinygrad.git
-python3 -m pip install -e ".[mlperf]"
-```
-Also install tqdm and tensorflow.
-```
-pip install tqdm tensorflow
-```
-
-### tinybox_green
-Install the p2p driver per [README](https://github.com/tinygrad/open-gpu-kernel-modules/blob/550.54.15-p2p/README.md)
-This is the default on production tinybox green.
-
-### tinybox_red
-Disable cwsr + increase mes timeout.
-Install the custom amdgpu driver per [README](https://github.com/nimlgen/amdgpu_ubuntu_22_04/blob/v6.1.3/readme.md)
-
-# 2. Directions
-
-## Steps to download and verify data
-
-### 1. Download raw data
-
-```
-BASEDIR="/raid/datasets/wiki" WIKI_TRAIN=1 VERIFY_CHECKSUM=1 python3 extra/datasets/wikipedia_download.py
-```
-
-### 2. Preprocess train and validation data
-
-Note: The number of threads used for preprocessing is limited by available memory. With 128GB of RAM, a maximum of 16 threads is recommended. 
-
-#### Training:
-```
-BASEDIR="/raid/datasets/wiki" NUM_WORKERS=16 python3 extra/datasets/wikipedia.py pre-train all
-```
-
-Generating a specific topic (Between 0 and 499)
-```
-BASEDIR="/raid/datasets/wiki" python3 extra/datasets/wikipedia.py pre-train 42
-```
-
-#### Validation:
-```
-BASEDIR="/raid/datasets/wiki" python3 extra/datasets/wikipedia.py pre-eval
-```
-## Running
-
-### tinybox_green
-
-#### Steps to run benchmark
-```
-examples/mlperf/training_submission_v4.1/tinycorp/benchmarks/bert/implementations/tinybox_green/run_and_time.sh
-```
-
-### tinybox_red
-
-#### One time setup
-
-```
-examples/mlperf/training_submission_v4.1/tinycorp/benchmarks/bert/implementations/tinybox_red/setup.sh
-```
-
-#### Steps to run benchmark
-```
-examples/mlperf/training_submission_v4.1/tinycorp/benchmarks/bert/implementations/tinybox_red/run_and_time.sh
-```
@@ -1,13 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="bert"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=72 EVAL_BS=6
-
-export BEAM=4 BEAM_UOPS_MAX=2000 BEAM_UPCAST_MAX=64 BEAM_LOCAL_MAX=512
-export IGNORE_JIT_FIRST_BEAM=1
-export BASEDIR="/raid/datasets/wiki"
-
-export BENCHMARK=10 DEBUG=2
-
-python3 examples/mlperf/model_train.py
@@ -1,13 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="bert"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=72 EVAL_BS=6
-
-export BEAM=4 BEAM_UOPS_MAX=2000 BEAM_UPCAST_MAX=64 BEAM_LOCAL_MAX=512
-export IGNORE_JIT_FIRST_BEAM=1
-export BASEDIR="/raid/datasets/wiki"
-
-export WANDB=1 PARALLEL=0
-
-RUNMLPERF=1 python3 examples/mlperf/model_train.py
@@ -1,23 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="bert"
-export SUBMISSION_PLATFORM="tinybox_green"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=72 EVAL_BS=6
-
-export BEAM=4 BEAM_UOPS_MAX=2000 BEAM_UPCAST_MAX=64 BEAM_LOCAL_MAX=512
-export IGNORE_JIT_FIRST_BEAM=1
-export BASEDIR="/raid/datasets/wiki"
-
-# pip install -e ".[mlperf]"
-export LOGMLPERF=1
-
-export SEED=$RANDOM
-DATETIME=$(date "+%m%d%H%M")
-LOGFILE="bert_green_${DATETIME}_${SEED}.log"
-
-# init
-BENCHMARK=10 INITMLPERF=1 python3 examples/mlperf/model_train.py | tee $LOGFILE
-
-# run
-PARALLEL=0 RUNMLPERF=1 python3 examples/mlperf/model_train.py | tee -a $LOGFILE
@@ -1,73 +0,0 @@
-# 1. Problem
-
-This problem uses BERT for NLP.
-
-## Requirements
-
-Install tinygrad and mlperf-logging from master.
-```
-git clone https://github.com/tinygrad/tinygrad.git
-python3 -m pip install -e ".[mlperf]"
-```
-Also install tqdm and tensorflow.
-```
-pip install tqdm tensorflow
-```
-
-### tinybox_green
-Install the p2p driver per [README](https://github.com/tinygrad/open-gpu-kernel-modules/blob/550.54.15-p2p/README.md)
-This is the default on production tinybox green.
-
-### tinybox_red
-Disable cwsr + increase mes timeout.
-Install the custom amdgpu driver per [README](https://github.com/nimlgen/amdgpu_ubuntu_22_04/blob/v6.1.3/readme.md)
-
-# 2. Directions
-
-## Steps to download and verify data
-
-### 1. Download raw data
-
-```
-BASEDIR="/raid/datasets/wiki" WIKI_TRAIN=1 VERIFY_CHECKSUM=1 python3 extra/datasets/wikipedia_download.py
-```
-
-### 2. Preprocess train and validation data
-
-Note: The number of threads used for preprocessing is limited by available memory. With 128GB of RAM, a maximum of 16 threads is recommended. 
-
-#### Training:
-```
-BASEDIR="/raid/datasets/wiki" NUM_WORKERS=16 python3 extra/datasets/wikipedia.py pre-train all
-```
-
-Generating a specific topic (Between 0 and 499)
-```
-BASEDIR="/raid/datasets/wiki" python3 extra/datasets/wikipedia.py pre-train 42
-```
-
-#### Validation:
-```
-BASEDIR="/raid/datasets/wiki" python3 extra/datasets/wikipedia.py pre-eval
-```
-## Running
-
-### tinybox_green
-
-#### Steps to run benchmark
-```
-examples/mlperf/training_submission_v4.1/tinycorp/benchmarks/bert/implementations/tinybox_green/run_and_time.sh
-```
-
-### tinybox_red
-
-#### One time setup
-
-```
-examples/mlperf/training_submission_v4.1/tinycorp/benchmarks/bert/implementations/tinybox_red/setup.sh
-```
-
-#### Steps to run benchmark
-```
-examples/mlperf/training_submission_v4.1/tinycorp/benchmarks/bert/implementations/tinybox_red/run_and_time.sh
-```
@@ -1,13 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="bert"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=72 EVAL_BS=6
-
-export BEAM=3
-export IGNORE_JIT_FIRST_BEAM=1
-export BASEDIR="/raid/datasets/wiki"
-
-export BENCHMARK=10 DEBUG=2
-
-python3 examples/mlperf/model_train.py
@@ -1,13 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="bert"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=72 EVAL_BS=6
-
-export BEAM=3
-export IGNORE_JIT_FIRST_BEAM=1
-export BASEDIR="/raid/datasets/wiki"
-
-export WANDB=1 PARALLEL=0
-
-RUNMLPERF=1 python3 examples/mlperf/model_train.py
@@ -1,23 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="bert"
-export SUBMISSION_PLATFORM="tinybox_red"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=72 EVAL_BS=6
-
-export BEAM=3
-export IGNORE_JIT_FIRST_BEAM=1
-export BASEDIR="/raid/datasets/wiki"
-
-# pip install -e ".[mlperf]"
-export LOGMLPERF=1
-
-export SEED=$RANDOM
-DATETIME=$(date "+%m%d%H%M")
-LOGFILE="bert_red_${DATETIME}_${SEED}.log"
-
-# init
-BENCHMARK=10 INITMLPERF=1 python3 examples/mlperf/model_train.py | tee $LOGFILE
-
-# run
-PARALLEL=0 RUNMLPERF=1 python3 examples/mlperf/model_train.py | tee -a $LOGFILE
@@ -1,8 +0,0 @@
-#!/bin/bash
-
-rocm-smi --setprofile compute
-rocm-smi --setmclk 3
-rocm-smi --setperflevel high
-
-# power cap to 350W
-# echo "350000000" | sudo tee /sys/class/drm/card{1..6}/device/hwmon/hwmon*/power1_cap
@@ -1,50 +0,0 @@
-# 1. Problem
-
-This problem uses the ResNet-50 CNN to do image classification.
-
-## Requirements
-
-Install tinygrad and mlperf-logging from master.
-```
-git clone https://github.com/tinygrad/tinygrad.git
-python3 -m pip install -e ".[mlperf]"
-```
-
-### tinybox_green
-Install the p2p driver per [README](https://github.com/tinygrad/open-gpu-kernel-modules/blob/550.54.15-p2p/README.md)
-This is the default on production tinybox green.
-
-### tinybox_red
-Disable cwsr
-This is the default on production tinybox red.
-```
-sudo vi /etc/modprobe.d/amdgpu.conf
-cat <<EOF > /etc/modprobe.d/amdgpu.conf
-options amdgpu cwsr_enable=0
-EOF
-sudo update-initramfs -u
-sudo reboot
-
-# validate
-sudo cat /sys/module/amdgpu/parameters/cwsr_enable #= 0
-```
-
-# 2. Directions
-
-## Steps to download and verify data
-
-```
-IMGNET_TRAIN=1 python3 extra/datasets/imagenet_download.py
-```
-
-## Steps for one time setup
-
-### tinybox_red
-```
-examples/mlperf/training_submission_v4.0/tinycorp/benchmarks/resnet/implementations/tinybox_red/setup.sh
-```
-
-## Steps to run benchmark
-```
-examples/mlperf/training_submission_v4.0/tinycorp/benchmarks/resnet/implementations/tinybox_red/run_and_time.sh
-```
@@ -1,13 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="resnet"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=1536 EVAL_BS=192
-
-export RESET_STEP=0
-
-export TRAIN_BEAM=4 IGNORE_JIT_FIRST_BEAM=1 BEAM_UOPS_MAX=1500 BEAM_UPCAST_MAX=64 BEAM_LOCAL_MAX=1024 BEAM_MIN_PROGRESS=10 BEAM_PADTO=0
-
-export BENCHMARK=10 DEBUG=2
-
-python3 examples/mlperf/model_train.py
@@ -1,15 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="resnet"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=1536 EVAL_BS=192
-
-export RESET_STEP=0
-
-export TRAIN_BEAM=4 IGNORE_JIT_FIRST_BEAM=1 BEAM_UOPS_MAX=1500 BEAM_UPCAST_MAX=64 BEAM_LOCAL_MAX=1024 BEAM_MIN_PROGRESS=10 BEAM_PADTO=0
-
-export EVAL_START_EPOCH=3 EVAL_FREQ=4
-
-export WANDB=1 PARALLEL=0
-
-python3 examples/mlperf/model_train.py
@@ -1,23 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="resnet"
-export SUBMISSION_PLATFORM="tinybox_green"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=1536 EVAL_BS=192
-
-export RESET_STEP=0
-
-export TRAIN_BEAM=4 IGNORE_JIT_FIRST_BEAM=1 BEAM_UOPS_MAX=1500 BEAM_UPCAST_MAX=64 BEAM_LOCAL_MAX=1024 BEAM_MIN_PROGRESS=10 BEAM_PADTO=0
-
-# pip install -e ".[mlperf]"
-export LOGMLPERF=1
-
-export SEED=$RANDOM
-DATETIME=$(date "+%m%d%H%M")
-LOGFILE="resnet_green_${DATETIME}_${SEED}.log"
-
-# init
-BENCHMARK=10 INITMLPERF=1 python3 examples/mlperf/model_train.py | tee $LOGFILE
-
-# run
-PARALLEL=0 RUNMLPERF=1 EVAL_START_EPOCH=3 EVAL_FREQ=4 python3 examples/mlperf/model_train.py | tee -a $LOGFILE
@@ -1,50 +0,0 @@
-# 1. Problem
-
-This problem uses the ResNet-50 CNN to do image classification.
-
-## Requirements
-
-Install tinygrad and mlperf-logging from master.
-```
-git clone https://github.com/tinygrad/tinygrad.git
-python3 -m pip install -e ".[mlperf]"
-```
-
-### tinybox_green
-Install the p2p driver per [README](https://github.com/tinygrad/open-gpu-kernel-modules/blob/550.54.15-p2p/README.md)
-This is the default on production tinybox green.
-
-### tinybox_red
-Disable cwsr
-This is the default on production tinybox red.
-```
-sudo vi /etc/modprobe.d/amdgpu.conf
-cat <<EOF > /etc/modprobe.d/amdgpu.conf
-options amdgpu cwsr_enable=0
-EOF
-sudo update-initramfs -u
-sudo reboot
-
-# validate
-sudo cat /sys/module/amdgpu/parameters/cwsr_enable #= 0
-```
-
-# 2. Directions
-
-## Steps to download and verify data
-
-```
-IMGNET_TRAIN=1 python3 extra/datasets/imagenet_download.py
-```
-
-## Steps for one time setup
-
-### tinybox_red
-```
-examples/mlperf/training_submission_v4.0/tinycorp/benchmarks/resnet/implementations/tinybox_red/setup.sh
-```
-
-## Steps to run benchmark
-```
-examples/mlperf/training_submission_v4.0/tinycorp/benchmarks/resnet/implementations/tinybox_red/run_and_time.sh
-```
@@ -1,13 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="resnet"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=1536 EVAL_BS=192
-
-export RESET_STEP=0
-
-export TRAIN_BEAM=4 IGNORE_JIT_FIRST_BEAM=1 BEAM_UOPS_MAX=2000 BEAM_UPCAST_MAX=96 BEAM_LOCAL_MAX=1024 BEAM_MIN_PROGRESS=5 BEAM_PADTO=0
-
-export BENCHMARK=10 DEBUG=2
-
-python3 examples/mlperf/model_train.py
@@ -1,15 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="resnet"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=1536 EVAL_BS=192
-
-export RESET_STEP=0
-
-export TRAIN_BEAM=4 IGNORE_JIT_FIRST_BEAM=1 BEAM_UOPS_MAX=2000 BEAM_UPCAST_MAX=96 BEAM_LOCAL_MAX=1024 BEAM_MIN_PROGRESS=5 BEAM_PADTO=0
-
-export EVAL_START_EPOCH=3 EVAL_FREQ=4
-
-export WANDB=1 PARALLEL=0
-
-python3 examples/mlperf/model_train.py
@@ -1,23 +0,0 @@
-#!/bin/bash
-
-export PYTHONPATH="."
-export MODEL="resnet"
-export SUBMISSION_PLATFORM="tinybox_red"
-export DEFAULT_FLOAT="HALF" GPUS=6 BS=1536 EVAL_BS=192
-
-export RESET_STEP=0
-
-export TRAIN_BEAM=4 IGNORE_JIT_FIRST_BEAM=1 BEAM_UOPS_MAX=2000 BEAM_UPCAST_MAX=96 BEAM_LOCAL_MAX=1024 BEAM_MIN_PROGRESS=5 BEAM_PADTO=0
-
-# pip install -e ".[mlperf]"
-export LOGMLPERF=1
-
-export SEED=$RANDOM
-DATETIME=$(date "+%m%d%H%M")
-LOGFILE="resnet_red_${DATETIME}_${SEED}.log"
-
-# init
-BENCHMARK=10 INITMLPERF=1 python3 examples/mlperf/model_train.py | tee $LOGFILE
-
-# run
-PARALLEL=0 RUNMLPERF=1 EVAL_START_EPOCH=3 EVAL_FREQ=4 python3 examples/mlperf/model_train.py | tee -a $LOGFILE
@@ -1,8 +0,0 @@
-#!/bin/bash
-
-rocm-smi --setprofile compute
-rocm-smi --setmclk 3
-rocm-smi --setperflevel high
-
-# power cap to 350W
-echo "350000000" | sudo tee /sys/class/drm/card{1..6}/device/hwmon/hwmon*/power1_cap
@@ -1,38 +0,0 @@
-{
-  "submitter": "tinycorp",
-  "division": "closed",
-  "status": "Available on-premise",
-  "system_name": "tinybox green",
-  "number_of_nodes": "1",
-  "host_processors_per_node": "1",
-  "host_processor_model_name": "AMD EPYC 7532 32-Core Processor",
-  "host_processor_core_count": "32",
-  "host_processor_vcpu_count": "64",
-  "host_processor_frequency": "",
-  "host_processor_caches": "",
-  "host_processor_interconnect": "",
-  "host_memory_capacity": "128GB",
-  "host_storage_type": "NVMe SSD",
-  "host_storage_capacity": "4 TB raid array + 1 TB boot",
-  "host_networking": "",
-  "host_networking_topology": "",
-  "host_memory_configuration": "8x 16GB DDR4",
-  "accelerators_per_node": "6",
-  "accelerator_model_name": "NVIDIA GeForce RTX 4090",
-  "accelerator_host_interconnect": "PCIe 4.0 x16",
-  "accelerator_frequency": "",
-  "accelerator_on-chip_memories": "",
-  "accelerator_memory_configuration": "GDDR6X",
-  "accelerator_memory_capacity": "24GB",
-  "accelerator_interconnect": "",
-  "accelerator_interconnect_topology": "",
-  "cooling": "air",
-  "hw_notes": "",
-  "framework": "tinygrad, commit b5546912e24e0a864b35924da4efa5d71cfe368b",
-  "other_software_stack": {
-    "python": "3.10.12",
-    "CUDA": "12.4"
-  },
-  "operating_system": "Ubuntu 22.04.4",
-  "sw_notes": ""
-}
@@ -1,38 +0,0 @@
-{
-  "submitter": "tinycorp",
-  "division": "closed",
-  "status": "Available on-premise",
-  "system_name": "tinybox red",
-  "number_of_nodes": "1",
-  "host_processors_per_node": "1",
-  "host_processor_model_name": "AMD EPYC 7532 32-Core Processor",
-  "host_processor_core_count": "32",
-  "host_processor_vcpu_count": "64",
-  "host_processor_frequency": "",
-  "host_processor_caches": "",
-  "host_processor_interconnect": "",
-  "host_memory_capacity": "128GB",
-  "host_storage_type": "NVMe SSD",
-  "host_storage_capacity": "4 TB raid array + 1 TB boot",
-  "host_networking": "",
-  "host_networking_topology": "",
-  "host_memory_configuration": "8x 16GB DDR4",
-  "accelerators_per_node": "6",
-  "accelerator_model_name": "AMD Radeon RX 7900 XTX",
-  "accelerator_host_interconnect": "PCIe 4.0 x16",
-  "accelerator_frequency": "",
-  "accelerator_on-chip_memories": "",
-  "accelerator_memory_configuration": "GDDR6",
-  "accelerator_memory_capacity": "24GB",
-  "accelerator_interconnect": "",
-  "accelerator_interconnect_topology": "",
-  "cooling": "air",
-  "hw_notes": "",
-  "framework": "tinygrad, commit b5546912e24e0a864b35924da4efa5d71cfe368b",
-  "other_software_stack": {
-    "python": "3.10.12",
-    "ROCm": "6.1.3"
-  },
-  "operating_system": "Ubuntu 22.04.4",
-  "sw_notes": ""
-}
@@ -0,0 +1,210 @@
+#!/usr/bin/env python3
+import os, sys, io, pathlib, json, struct
+import numpy as np
+sys.path.insert(0, str(pathlib.Path(__file__).parents[1]))
+
+if "FLOAT16" not in os.environ: os.environ["FLOAT16"] = "1"
+if "IMAGE" not in os.environ: os.environ["IMAGE"] = "2"
+if "NOLOCALS" not in os.environ: os.environ["NOLOCALS"] = "1"
+
+OPENPILOT_MODEL = "https://github.com/commaai/openpilot/raw/v0.9.4/selfdrive/modeld/models/supercombo.onnx"
+
+import onnx
+from typing import Tuple, List, Optional, Dict, cast
+from extra.onnx import get_run_onnx
+from tinygrad import Tensor, Device, GlobalCounters, dtypes
+from tinygrad.dtype import ImageDType
+from tinygrad.device import Buffer
+from tinygrad.helpers import partition, Context, fetch, getenv, DEBUG, tqdm
+from tinygrad.engine.realize import run_schedule, lower_schedule, ExecItem, CompiledRunner, memory_planner
+from tinygrad.engine.schedule import ScheduleItem, create_schedule
+from tinygrad.ops import UOps
+from tinygrad.tensor import _to_np_dtype
+Device.DEFAULT = "GPU"
+
+def get_schedule(onnx_data) -> Tuple[List[ScheduleItem], List[ScheduleItem]]:
+  Tensor.no_grad = True
+  Tensor.training = False
+
+  # load the model
+  onnx_model = onnx.load(io.BytesIO(onnx_data))
+  run_onnx = get_run_onnx(onnx_model)
+  input_shapes = {inp.name:tuple(x.dim_value for x in inp.type.tensor_type.shape.dim) for inp in onnx_model.graph.input}
+
+  # run the model
+  inputs = {k:Tensor.empty(*shp) for k,shp in input_shapes.items()}
+  ret: Tensor = next(iter(run_onnx(inputs).values())).cast(dtypes.float32).contiguous()
+  schedule = create_schedule([ret.lazydata])
+
+  # filter schedule that don't depend on the inputs
+  input_lb = [x.lazydata.base.buffer for x in inputs.values()]
+  depends = set(input_lb)
+  for si in schedule:
+    if any(b in depends for b in si.inputs):
+      for out in si.outputs: depends.add(out)
+
+  # run all kernels that don't depend on the inputs
+  # NOTE: there's two extra kernels due to fusions that now happen since the weights aren't realized
+  schedule, schedule_independent = partition(schedule, lambda si: any(out in depends for out in si.outputs))
+  print(f"{len(schedule)} schedule items depend on the input, {len(schedule_independent)} don't")
+
+  # confirm no non-sink metaop in the (non independent) schedule except for the ones that load the input buffers
+  assert all(si.ast.op is UOps.SINK or out in input_lb for si in schedule for out in si.outputs), "has non SINK ops, can't compile to Thneed"
+  return schedule, schedule_independent, inputs
+
+def test_vs_onnx(onnx_data, eis:Optional[List[ExecItem]], inputs:Dict[str, Tensor]):
+  import onnx
+  #import pyopencl as cl
+  #from extra.thneed import Thneed
+  import numpy as np
+  onnx_model = onnx.load(io.BytesIO(onnx_data))
+
+  input_shapes = {inp.name:tuple(x.dim_value for x in inp.type.tensor_type.shape.dim) for inp in onnx_model.graph.input}
+  Tensor.manual_seed(1337)
+  new_inputs = {k:Tensor.randn(*shp, requires_grad=False)*8 for k,shp in input_shapes.items()}
+  new_np_inputs = {k:v.realize().numpy() for k,v in new_inputs.items()}
+
+  if getenv("ORT"):
+    # test with onnxruntime
+    import onnxruntime as ort
+    onnx_session = ort.InferenceSession(onnx_data)
+    onnx_output = onnx_session.run([onnx_model.graph.output[0].name], {k:v.astype(np.float16) for k,v in new_np_inputs.items()})
+    new_torch_out = onnx_output[0]
+    print("got ort outputs")
+  else:
+    # test with torch
+    from test.models.test_onnx import run_onnx_torch
+    new_torch_out = run_onnx_torch(onnx_model, new_np_inputs).numpy()
+    print("got torch outputs")
+
+  # if you don't have a schedule
+  if eis is None:
+    run_onnx = get_run_onnx(onnx_model)
+    new_tinygrad_out = next(iter(run_onnx(new_inputs).values())).cast(dtypes.float32).numpy()
+    np.testing.assert_allclose(new_torch_out, new_tinygrad_out, atol=1e-4, rtol=1e-2)
+    print("classic self-test passed!")
+    return
+
+  # set inputs
+  for k,v in inputs.items(): v.lazydata.base.realized.copyin(new_np_inputs[k].data)
+
+  # run code (all buffers have been allocated)
+  GlobalCounters.reset()
+  output = eis[-1].bufs[0]
+  for ei in eis: ei.run()
+
+  new_tinygrad_out = np.frombuffer(output.as_buffer(), dtype=_to_np_dtype(output.dtype))
+  np.testing.assert_allclose(new_torch_out.reshape(new_tinygrad_out.shape), new_tinygrad_out, atol=1e-4, rtol=1e-2)
+  print("semi-thneed self-test passed!")
+
+if __name__ == "__main__":
+  onnx_data = fetch(sys.argv[1] if len(sys.argv) > 1 else OPENPILOT_MODEL).read_bytes()
+
+  # quick test for ONNX issues
+  #thneed_test_onnx(onnx_data, None)
+  #exit(0)
+
+  schedule, schedule_independent, inputs = get_schedule(onnx_data)
+  schedule, schedule_input = partition(schedule, lambda x: x.ast.op is UOps.SINK)
+  print(f"{len(schedule_input)} inputs")
+
+  run_schedule(schedule_independent)
+  run_schedule(schedule_input)
+  with Context(DEBUG=max(DEBUG.value, 2), BEAM=getenv("LATEBEAM")):
+    schedule = memory_planner(schedule)
+    for si in schedule:
+      for b in si.outputs:
+        assert not b.is_allocated(), "output should not be allocated"
+    image_count = sum(isinstance(out.dtype, ImageDType) for si in schedule for out in si.outputs)
+    print(f"**** compiling real kernels {image_count}/{len(schedule)} images ****")
+    eis = list(tqdm(lower_schedule(schedule), total=len(schedule)))
+
+  print("kernel count:", len(eis))
+  assert len(eis) <= getenv("ALLOWED_KERNEL_COUNT", 0) or getenv("ALLOWED_KERNEL_COUNT", 0) == 0, "too many kernels!"
+
+  # new simple thneed
+  def to_ref(b:Buffer): return struct.pack("Q", id(b)).decode("latin_1")
+
+  seen_buffers = set()
+  input_buffers = [x.lazydata.buffer for x in inputs.values()]
+  jdat = {"binaries": [], "programs": {}, "kernels": [], "objects": []}
+  jdat["inputs"] = {k:to_ref(v.lazydata.buffer) for k,v in inputs.items()}
+  jdat["outputs"] = [to_ref(eis[-1].bufs[0])]
+  weights = []
+  for i,ei in enumerate(eis):
+    #print("***", i)
+    for b in ei.bufs:
+      needs_load = b.is_allocated() and b not in input_buffers
+      #print(b, needs_load)
+      if b in seen_buffers: continue
+      seen_buffers.add(b)
+      if isinstance(b.dtype, ImageDType):
+        base_dtype = dtypes.float16 if b.dtype.fmt == 'e' else dtypes.float32
+        row_pitch = (b.dtype.shape[0]*4*base_dtype.itemsize + 63)//64 * 64
+        size = row_pitch * b.dtype.shape[1]
+        jdat['objects'].append({
+          "id": to_ref(b), "needs_load": needs_load, "size": size, "arg_type": "image2d_t",
+          "width": b.dtype.shape[0], "height": b.dtype.shape[1], "row_pitch": row_pitch, "float32": b.dtype.base == dtypes.float32,
+        })
+        if needs_load:
+          t = Tensor.empty(b.dtype.shape, dtype=b.dtype)
+          t.lazydata.buffer = b
+          data = t.cast(dtypes.float32).pad(((0, row_pitch//(4*base_dtype.itemsize)-b.dtype.shape[0]), (0,0), (0,0))).contiguous().numpy()
+          # NOTE: this cast must be done in numpy for platforms that don't support half
+          if base_dtype == dtypes.float16: data = data.astype(np.float16)
+          weights.append(data.tobytes())
+          assert len(weights[-1]) == size, "wrong size buffer"
+      else:
+        jdat['objects'].append({
+          "id": to_ref(b), "arg_type": b.dtype.name + "*", "needs_load": needs_load, "size": b.nbytes,
+        })
+        if needs_load:
+          weights.append(b.as_buffer())
+          assert len(weights[-1]) == b.nbytes, "wrong size buffer"
+
+  saved_binaries = set()
+  binaries = []
+  gated_read_image_count = 0
+  GlobalCounters.reset()
+  with Context(DEBUG=max(DEBUG.value, 2)):
+    for ei in eis:
+      prg = cast(CompiledRunner, ei.prg)
+      assert len(prg.p.vars) == 0
+      if prg.p.function_name not in saved_binaries:
+        jdat['binaries'].append({"name":prg.p.function_name, "length":len(prg.lib)})
+        binaries.append(prg.lib)
+        saved_binaries.add(prg.p.function_name)
+      gated_read_image_count += prg.p.src.count("?read_image")
+      ei.run()
+      jdat['kernels'].append({
+        "name": prg.p.function_name,
+        "work_dim": len(prg.p.global_size),
+        "global_work_size": prg.p.global_size,
+        "local_work_size": prg.p.local_size,
+        "num_args": len(ei.bufs),
+        "args": [to_ref(b) for b in ei.bufs],
+        "arg_size": [8]*len(ei.bufs),
+      })
+
+  if (allowed_gated_read_image:=getenv("ALLOWED_GATED_READ_IMAGE", -1)) != -1:
+    assert gated_read_image_count <= allowed_gated_read_image, \
+      f"too many gated read_image! {gated_read_image_count=}, {allowed_gated_read_image=}"
+
+  output_fn = sys.argv[2] if len(sys.argv) >= 3 else "/tmp/output.thneed"
+  print(f"saving thneed to {output_fn} with {len(weights)} buffers and {len(binaries)} binaries")
+  with open(output_fn, "wb") as f:
+    j = json.dumps(jdat, ensure_ascii=False).encode('latin_1')
+    f.write(struct.pack("I", len(j)))
+    f.write(j)
+    for w in weights: f.write(w)
+    for b in binaries: f.write(b)
+    print("saved", f.tell(), "bytes")
+
+  FLOAT16 = getenv("FLOAT16", 0)
+  if FLOAT16 == 0:
+    try:
+      test_vs_onnx(onnx_data, eis, inputs)
+    except ModuleNotFoundError as e:
+      print(f"TEST NOT HAPPENING {e}")
+
+
@@ -5,10 +5,9 @@ if "IMAGE" not in os.environ: os.environ["IMAGE"] = "2"
 if "NOLOCALS" not in os.environ: os.environ["NOLOCALS"] = "1"
 if "JIT_BATCH_SIZE" not in os.environ: os.environ["JIT_BATCH_SIZE"] = "0"

-from tinygrad import fetch, Tensor, TinyJit, Context, GlobalCounters, Device
-from tinygrad.helpers import DEBUG, getenv
+from tinygrad import fetch, Tensor, TinyJit, Device, Context, GlobalCounters
+from tinygrad.helpers import OSX, DEBUG, getenv
 from tinygrad.tensor import _from_np_dtype
-from tinygrad.engine.realize import CompiledRunner

 import onnx
 from onnx.helper import tensor_dtype_to_np_dtype
@@ -17,120 +16,65 @@ from extra.onnx import get_run_onnx   # TODO: port to main tinygrad
 OPENPILOT_MODEL = sys.argv[1] if len(sys.argv) > 1 else "https://github.com/commaai/openpilot/raw/v0.9.7/selfdrive/modeld/models/supercombo.onnx"
 OUTPUT = "/tmp/openpilot.pkl"

-def compile(onnx_file):
-  onnx_model = onnx.load(onnx_file)
+def compile():
+  # hack to fix GPU on OSX: max doesn't work on half, see test/external/external_gpu_fail_osx.py
+  if OSX:
+    from tinygrad.ops import BinaryOps
+    from tinygrad.renderer.cstyle import ClangRenderer, CStyleLanguage
+    CStyleLanguage.code_for_op[BinaryOps.MAX] = ClangRenderer.code_for_op[BinaryOps.MAX]
+
  Tensor.no_grad = True
  Tensor.training = False

+  onnx_bytes = fetch(OPENPILOT_MODEL)
+  onnx_model = onnx.load(onnx_bytes)
  run_onnx = get_run_onnx(onnx_model)
  print("loaded model")

  input_shapes = {inp.name:tuple(x.dim_value for x in inp.type.tensor_type.shape.dim) for inp in onnx_model.graph.input}
  input_types = {inp.name: tensor_dtype_to_np_dtype(inp.type.tensor_type.elem_type) for inp in onnx_model.graph.input}
-  if getenv("FLOAT16", 0) == 0: input_types = {k:(np.float32 if v==np.float16 else v) for k,v in input_types.items()}
  Tensor.manual_seed(100)
  new_inputs = {k:Tensor.randn(*shp, dtype=_from_np_dtype(input_types[k])).mul(8).realize() for k,shp in sorted(input_shapes.items())}
-  new_inputs_numpy = {k:v.numpy() for k,v in new_inputs.items()}
  print("created tensors")

-  run_onnx_jit = TinyJit(lambda **kwargs:
-                         next(iter(run_onnx({k:v.to(Device.DEFAULT) for k,v in kwargs.items()}).values())).cast('float32'), prune=True)
+  run_onnx_jit = TinyJit(lambda **kwargs: run_onnx(kwargs), prune=True)
  for i in range(3):
    GlobalCounters.reset()
    print(f"run {i}")
-    inputs = {**{k:v.clone() for k,v in new_inputs.items() if 'img' in k},
-              **{k:Tensor(v, device="NPY").realize() for k,v in new_inputs_numpy.items() if 'img' not in k}}
    with Context(DEBUG=max(DEBUG.value, 2 if i == 2 else 1)):
-      ret = run_onnx_jit(**inputs).numpy()
-    # copy i == 1 so use of JITBEAM is okay
-    if i == 1: test_val = np.copy(ret)
+      ret = next(iter(run_onnx_jit(**new_inputs).values())).cast('float32').numpy()
+    if i == 0: test_val = np.copy(ret)
  print(f"captured {len(run_onnx_jit.captured.jit_cache)} kernels")
-  np.testing.assert_equal(test_val, ret, "JIT run failed")
+  np.testing.assert_equal(test_val, ret)
  print("jit run validated")

-  # checks from compile2
-  kernel_count = 0
-  read_image_count = 0
-  gated_read_image_count = 0
-  for ei in run_onnx_jit.captured.jit_cache:
-    if isinstance(ei.prg, CompiledRunner):
-      kernel_count += 1
-      read_image_count += ei.prg.p.src.count("read_image")
-      gated_read_image_count += ei.prg.p.src.count("?read_image")
-  print(f"{kernel_count=},  {read_image_count=}, {gated_read_image_count=}")
-  if (allowed_kernel_count:=getenv("ALLOWED_KERNEL_COUNT", -1)) != -1:
-    assert kernel_count <= allowed_kernel_count, f"too many kernels! {kernel_count=}, {allowed_kernel_count=}"
-  if (allowed_read_image:=getenv("ALLOWED_READ_IMAGE", -1)) != -1:
-    assert read_image_count == allowed_read_image, f"different read_image! {read_image_count=}, {allowed_read_image=}"
-  if (allowed_gated_read_image:=getenv("ALLOWED_GATED_READ_IMAGE", -1)) != -1:
-    assert gated_read_image_count <= allowed_gated_read_image, f"too many gated read_image! {gated_read_image_count=}, {allowed_gated_read_image=}"
-
  with open(OUTPUT, "wb") as f:
    pickle.dump(run_onnx_jit, f)
-  mdl_sz = os.path.getsize(onnx_file)
+  mdl_sz = os.path.getsize(onnx_bytes)
  pkl_sz = os.path.getsize(OUTPUT)
  print(f"mdl size is {mdl_sz/1e6:.2f}M")
  print(f"pkl size is {pkl_sz/1e6:.2f}M")
  print("**** compile done ****")
  return test_val

-def test_vs_compile(run, new_inputs, test_val=None):
-  new_inputs_numpy = {k:v.numpy() for k,v in new_inputs.items()}
-
-  # create fake "from_blob" tensors for the inputs, and wrapped NPY tensors for the numpy inputs (these have the same underlying memory)
-  inputs = {**{k:v for k,v in new_inputs.items() if 'img' in k},
-            **{k:Tensor(v, device="NPY").realize() for k,v in new_inputs_numpy.items() if 'img' not in k}}
-
-  # run 20 times
+def test(test_val=None):
+  with open(OUTPUT, "rb") as f:
+    run = pickle.load(f)
+  Tensor.manual_seed(100)
+  new_inputs = {nm:Tensor.randn(*st.shape, dtype=dtype).mul(8).realize() for nm, (st, _, dtype, _) in
+                sorted(zip(run.captured.expected_names, run.captured.expected_st_vars_dtype_device))}
  for _ in range(20):
    st = time.perf_counter()
-    out = run(**inputs)
+    out = run(**new_inputs)
    mt = time.perf_counter()
-    val = out.numpy()
+    val = out['outputs'].numpy()
    et = time.perf_counter()
    print(f"enqueue {(mt-st)*1e3:6.2f} ms -- total run {(et-st)*1e3:6.2f} ms")
  print(out, val.shape, val.dtype)
  if test_val is not None: np.testing.assert_equal(test_val, val)
  print("**** test done ****")

-  # test that changing the numpy changes the model outputs
-  for v in new_inputs_numpy.values(): v *= 2
-  out = run(**inputs)
-  changed_val = out.numpy()
-  np.testing.assert_raises(AssertionError, np.testing.assert_array_equal, val, changed_val)
-  return val
-
-def test_vs_onnx(new_inputs, test_val, onnx_file):
-  new_inputs_numpy = {k:v.numpy() for k,v in new_inputs.items()}
-  onnx_model = onnx.load(onnx_file)
-
-  if getenv("ORT"):
-    # test with onnxruntime
-    import onnxruntime as ort
-    onnx_session = ort.InferenceSession(onnx_file)
-    onnx_output = onnx_session.run([onnx_model.graph.output[0].name], {k:v.astype(np.float16) for k,v in new_inputs_numpy.items()})
-    new_torch_out = onnx_output[0]
-    print("got ort outputs")
-  else:
-    # test with torch
-    from test.models.test_onnx import run_onnx_torch
-    # NOTE: we have to correct the order here
-    new_torch_out = run_onnx_torch(onnx_model, {k.name:new_inputs_numpy[k.name] for k in onnx_model.graph.input}).numpy()
-    print("got torch outputs")
-
-  np.testing.assert_allclose(new_torch_out.reshape(test_val.shape), test_val, atol=1e-4, rtol=1e-2)
-  print("test vs onnx passed")
-
 if __name__ == "__main__":
-  onnx_file = fetch(OPENPILOT_MODEL)
-  test_val = compile(onnx_file) if not getenv("RUN") else None
+  test_val = compile() if not getenv("RUN") else None
+  test(test_val)

-  with open(OUTPUT, "rb") as f: pickle_loaded = pickle.load(f)
-
-  # same randomness as compile
-  Tensor.manual_seed(100)
-  new_inputs = {nm:Tensor.randn(*st.shape, dtype=dtype).mul(8).realize() for nm, (st, _, dtype, _) in
-                sorted(zip(pickle_loaded.captured.expected_names, pickle_loaded.captured.expected_st_vars_dtype_device))}
-
-  test_val = test_vs_compile(pickle_loaded, new_inputs, test_val)
-  if not getenv("FLOAT16"): test_vs_onnx(new_inputs, test_val, onnx_file)
@@ -0,0 +1,2 @@
+#!/bin/bash
+NOLOCALS=1 FLOAT16=1 DEBUGCL=1 IMAGE=2 GPU=1 python3 examples/openpilot/compile2.py
@@ -1,117 +0,0 @@
-import argparse
-import os
-import sys
-
-from transformers import AutoTokenizer
-from pathlib import Path
-from typing import Dict, Union
-
-from extra.models.llama import Transformer, convert_from_huggingface, fix_bf16
-from examples.llama3 import load
-from tinygrad import nn, Tensor
-from tinygrad.helpers import fetch, colored, GlobalCounters, Timing, DEBUG
-from tinygrad.nn.state import load_state_dict, get_parameters
-
-MODELS = {
-  "32B": {
-    "model_params": {"dim": 5120, "n_heads": 40, "n_kv_heads": 8, "n_layers": 64, "norm_eps": 1e-5, "rope_theta": 1000000, "vocab_size": 152064, "hidden_dim": 27648},
-    "total_num_weights": 17,
-    "tokenizer": "Qwen/QwQ-32B-Preview"
-  }
-}
-
-def download_weights(total_num_weights:int) -> Path:
-  model = fetch("https://huggingface.co/Qwen/QwQ-32B-Preview/resolve/main/model.safetensors.index.json?download=true", "model.safetensors.index.json", subdir=(subdir:="qwq_32b_preview"))
-
-  for i in range(1, total_num_weights + 1):
-    filename = f"model-{i:05d}-of-{total_num_weights:05d}.safetensors"
-    fetch(f"https://huggingface.co/Qwen/QwQ-32B-Preview/resolve/main/{filename}?download=true", filename, subdir=subdir)
-
-  return Path(os.path.dirname(model))
-
-def load_model(model_path:Path, model_params:Dict[str, Union[int, float]]) -> Transformer:
-  # build model
-  model = Transformer(**model_params, linear=nn.Linear)
-
-  # update layers to add bias
-  updated_layers = []
-  for layer in model.layers:
-    head_dim = model_params["dim"] // model_params["n_heads"]
-    layer.attention.wq = nn.Linear(model_params["dim"], model_params["n_heads"] * head_dim, bias=True)
-    layer.attention.wk = nn.Linear(model_params["dim"], model_params["n_kv_heads"] * head_dim, bias=True)
-    layer.attention.wv = nn.Linear(model_params["dim"], model_params["n_kv_heads"] * head_dim, bias=True)
-    updated_layers.append(layer)
-  model.layers = updated_layers
-
-  # load weights
-  weights = fix_bf16(convert_from_huggingface(load(str(model_path / "model.safetensors.index.json")), model, model_params["n_heads"], model_params["n_kv_heads"], permute_layers=False))
-
-  # replace weights in model
-  load_state_dict(model, weights, strict=False, consume=True)
-  return model
-
-
-if __name__ == "__main__":
-  Tensor.no_grad = True
-
-  parser = argparse.ArgumentParser(description="Run QwQ in tinygrad", formatter_class=argparse.ArgumentDefaultsHelpFormatter)
-  parser.add_argument("--size", choices=["32B"], default="32B", help="Model size")
-  parser.add_argument("--count", type=int, default=30, help="Max number of tokens to generate")
-  parser.add_argument("--temperature", type=float, default=0.7, help="Temperature in the softmax")
-  parser.add_argument("--prompt", type=str, default="Hello.", help="Phrase to start with")
-  parser.add_argument("--weights", type=str, default=None, help="Path to the downloaded weights")
-  parser.add_argument("--timing", action="store_true", help="Print timing per token")
-  args = parser.parse_args()
-
-  model_info = MODELS[args.size]
-
-  model_path = Path(args.weights) if args.weights else download_weights(model_info["total_num_weights"])
-  transformer = load_model(model_path, model_info["model_params"])
-  tokenizer = AutoTokenizer.from_pretrained(model_info["tokenizer"])
-  param_bytes = sum(x.lazydata.size * x.dtype.itemsize for x in get_parameters(transformer))
-
-  outputted = args.prompt
-  start_pos, toks = 0, tokenizer(outputted)["input_ids"]
-  print(outputted, end="", flush=True)
-
-  tok_tensor = None
-  for i in range(args.count):
-    GlobalCounters.reset()
-
-    if args.timing: print("")
-    st = GlobalCounters.time_sum_s
-    next_tok = Tensor([toks[start_pos:]]) if tok_tensor is None or (len(toks)-start_pos) > 1 else tok_tensor.reshape(1, 1)
-    with Timing("total ", enabled=args.timing, on_exit=lambda x: f", {1e9/x:.2f} tok/s, {GlobalCounters.global_mem/x:.2f} GB/s, param {param_bytes/x:.2f} GB/s"):
-      with Timing("enqueue in ", on_exit=(lambda et: (f", {(GlobalCounters.time_sum_s-st)*1e3:.2f} ms on GPU" if DEBUG>=2 else "") +
-                  f", {GlobalCounters.global_ops*1e-9:.2f} GOPS, {GlobalCounters.global_mem*1e-9:.2f} GB" +
-                  (f", {GlobalCounters.global_mem*1e-9/(GlobalCounters.time_sum_s-st):.2f} GB/s, param {param_bytes*1e-9/(GlobalCounters.time_sum_s-st):.2f} GB/s" if DEBUG>=2 else "")) if DEBUG else None, enabled=args.timing):
-        tok_tensor = transformer(next_tok, start_pos, args.temperature)
-      tok = tok_tensor.item()
-
-    # use the kv cache
-    start_pos = len(toks)
-
-    # add the new token
-    toks.append(tok)
-
-    cur = tokenizer.decode(toks, skip_special_tokens=True)
-    sys.stdout.write(cur[len(outputted):])
-    sys.stdout.flush()
-    outputted = cur
-
-  if args.temperature == 0:
-    text = tokenizer.decode(toks)
-    key = (args.size, args.count, args.prompt)
-    expected = {
-      ("32B", 10, "Hello."): "Hello. I'm trying to make a program that will read",
-      (
-        "32B",
-        50,
-        "Can you tell me more about machine learning?"
-      ): "Can you tell me more about machine learning? Sure, I'd be happy to help! Machine learning is a subset of artificial intelligence that focuses on building systems that can learn from data and make predictions or decisions without being explicitly programmed to do so. It's a fascinating field with a lot of real"
-    }
-    try:
-      assert text == expected[key], f"invalid output: `{colored(text, 'red')}` != `{expected[key]}`"
-      print("\n" + colored("output validated", "green"))
-    except KeyError:
-      pass
@@ -303,7 +303,7 @@ class VanillaCFG(Guider):

 class SplitVanillaCFG(Guider):
  def __call__(self, denoiser, x:Tensor, s:Tensor, c:Dict, uc:Dict) -> Tensor:
-    x_u = denoiser(x, s, uc).clone().realize()
+    x_u = denoiser(x, s, uc)
    x_c = denoiser(x, s, c)
    x_pred = x_u + self.scale*(x_c - x_u)
    return x_pred
@@ -1,35 +0,0 @@
-import os, pathlib
-from examples.llama3 import Tokenizer
-from tabulate import tabulate
-from tinygrad import fetch
-from tinygrad.helpers import flatten
-
-# llama 3 tokenizer
-tokenizer = Tokenizer(fetch("https://huggingface.co/bofenghuang/Meta-Llama-3-8B/resolve/main/original/tokenizer.model").as_posix())
-
-def read_code(base_path):
-  ret = []
-  for path, _, files in os.walk(os.path.join(base_path, "tinygrad")):
-    for name in files:
-      if not name.endswith(".py"): continue
-      if 'tinygrad/runtime/autogen' in path.replace('\\', '/'): continue
-      fullpath = os.path.join(path, name)
-      code = pathlib.Path(fullpath).read_text()
-      ret += [(fullpath.split("tinygrad/", 1)[1], code)]
-  return ret
-
-if __name__ == "__main__":
-  ret = read_code(".")
-
-  table = []
-  for name,code in ret:
-    table.append([name, len(tokenizer.encode(name+"\x00"+code))])
-  print(tabulate([["name", "llm tokens"]]+sorted(table, key=lambda x: -x[1]), headers="firstrow"))
-
-  code_str = '\x00'.join(flatten(ret))
-  print(f"code has {len(code_str)} chars")
-  newline_count = code_str.count('\n')
-  print(f"code has {newline_count} newlines")
-
-  encoded = tokenizer.encode(code_str)
-  print(f"code has {len(encoded)} tokens")
@@ -45,7 +45,7 @@ class ConvBlock:
  def __call__(self, input):
    x = input.reshape(shape=(-1, self.inp, self.w, self.h))
    for cweight, cbias in zip(self.cweights, self.cbiases):
-      x = x.pad(padding=[1,1,1,1]).conv2d(cweight).add(cbias).relu()
+      x = x.pad2d(padding=[1,1,1,1]).conv2d(cweight).add(cbias).relu()
    x = self._bn(x)
    x = self._seb(x)
    return x
@@ -361,21 +361,21 @@ class SineGen:
    self.dim = self.harmonic_num + 1
  def _f02uv(self, f0): return (f0 > self.voiced_threshold).float()  #generate uv signal
  def _f02sine(self, f0_values):
-    def padDiff(x : Tensor): return (x.pad((0,0,-1,1)) - x).pad((0,0,0,-1))
+    def padDiff(x : Tensor): return (x.pad2d((0,0,-1,1)) - x).pad2d((0,0,0,-1))
    def mod(x: Tensor, n: int) -> Tensor: return x - n * x.div(n).floor()  # this is what the % operator does in pytorch.
    rad_values = mod((f0_values / self.sampling_rate) , 1)  # convert to F0 in rad
    rand_ini = Tensor.rand(f0_values.shape[0], f0_values.shape[2], device=f0_values.device)  # initial phase noise

    #rand_ini[:, 0] = 0
-    m = Tensor.ones(f0_values.shape[0]).unsqueeze(1).pad((0,f0_values.shape[2]-1,0,0)).cast(dtypes.bool)
+    m = Tensor.ones(f0_values.shape[0]).unsqueeze(1).pad2d((0,f0_values.shape[2]-1,0,0)).cast(dtypes.bool)
    m = tilde(m)
    rand_ini = m.where(rand_ini, 0)

    #rad_values[:, 0, :] = rad_values[:, 0, :] + rand_ini
    tmp = rad_values[:, 0, :] + rand_ini
-    m = Tensor.ones(tmp.shape).pad((0,0,0,rad_values.shape[1]-1,0)).cast(dtypes.bool)
+    m = Tensor.ones(tmp.shape).pad2d((0,0,0,rad_values.shape[1]-1,0)).cast(dtypes.bool)
    m = tilde(m)
-    tmp = tmp.unsqueeze(1).pad((0,0,0,rad_values.shape[1]-1,0))
+    tmp = tmp.unsqueeze(1).pad2d((0,0,0,rad_values.shape[1]-1,0))
    rad_values = m.where(rad_values, tmp)

    tmp_over_one = mod(rad_values.cumsum(1), 1)
@@ -383,7 +383,7 @@ class SineGen:
    cumsum_shift = Tensor.zeros_like(rad_values)

    #cumsum_shift[:, 1:, :] = tmp_over_one_idx * -1.0
-    tmp_over_one_idx = (tmp_over_one_idx * -1.0).pad((0,0,1,0))
+    tmp_over_one_idx = (tmp_over_one_idx * -1.0).pad2d((0,0,1,0))
    cumsum_shift = tmp_over_one_idx

    sines = ((rad_values + cumsum_shift).cumsum(1) * 2 * np.pi).sin()
@@ -137,7 +137,7 @@ class Resample:
    waveform = waveform.reshape(-1, shape[-1])  # pack batch
    num_wavs, length = waveform.shape
    target_length = int(math.ceil(new_freq * length / orig_freq))
-    waveform = waveform.pad((self.width, self.width + orig_freq))
+    waveform = waveform.pad2d((self.width, self.width + orig_freq))
    resampled = waveform[:, None].conv2d(self.kernel, stride=orig_freq)
    resampled = resampled.transpose(1, 2).reshape(num_wavs, -1)
    resampled = resampled[..., :target_length]
@@ -189,7 +189,7 @@ class StableDiffusion:
    # make image correct size and scale
    x = (x + 1.0) / 2.0
    x = x.reshape(3,512,512).permute(1,2,0).clip(0,1)*255
-    return x.cast(dtypes.uint8)
+    return x.cast(dtypes.uint8) if Device.DEFAULT != "WEBGPU" else x

  def __call__(self, unconditional_context, context, latent, timestep, alphas, alphas_prev, guidance):
    e_t = self.get_model_output(unconditional_context, context, latent, timestep, guidance)
@@ -280,7 +280,7 @@ if __name__ == "__main__":
  print(x.shape)

  # save image
-  im = Image.fromarray(x.numpy())
+  im = Image.fromarray(x.numpy().astype(np.uint8, copy=False))
  print(f"saving {args.out}")
  im.save(args.out)
  # Open image.
@@ -290,5 +290,5 @@ if __name__ == "__main__":
  if args.prompt == default_prompt and args.steps == 6 and args.seed == 0 and args.guidance == 7.5:
    ref_image = Tensor(np.array(Image.open(Path(__file__).parent / "stable_diffusion_seed0.png")))
    distance = (((x.cast(dtypes.float) - ref_image.cast(dtypes.float)) / ref_image.max())**2).mean().item()
-    assert distance < 3e-3, colored(f"validation failed with {distance=}", "red")  # higher distance with WINO
+    assert distance < 50e-5, colored(f"validation failed with {distance=}", "red")
    print(colored(f"output validated with {distance=}", "green"))
@@ -1 +0,0 @@
-(()=>{function o(e){e.directive("intersect",e.skipDuringClone((t,{value:i,expression:l,modifiers:n},{evaluateLater:r,cleanup:c})=>{let s=r(l),a={rootMargin:x(n),threshold:f(n)},u=new IntersectionObserver(d=>{d.forEach(h=>{h.isIntersecting!==(i==="leave")&&(s(),n.includes("once")&&u.disconnect())})},a);u.observe(t),c(()=>{u.disconnect()})}))}function f(e){if(e.includes("full"))return .99;if(e.includes("half"))return .5;if(!e.includes("threshold"))return 0;let t=e[e.indexOf("threshold")+1];return t==="100"?1:t==="0"?0:Number(`.${t}`)}function p(e){let t=e.match(/^(-?[0-9]+)(px|%)?$/);return t?t[1]+(t[2]||"px"):void 0}function x(e){let t="margin",i="0px 0px 0px 0px",l=e.indexOf(t);if(l===-1)return i;let n=[];for(let r=1;r<5;r++)n.push(p(e[l+r]||""));return n=n.filter(r=>r!==void 0),n.length?n.join(" ").trim():i}document.addEventListener("alpine:init",()=>{window.Alpine.plugin(o)});})();
@@ -1 +0,0 @@
-pre code.hljs{display:block;overflow-x:auto;padding:1em}code.hljs{padding:3px 5px}.hljs{background:#1e1e1e;color:#dcdcdc}.hljs-keyword,.hljs-literal,.hljs-name,.hljs-symbol{color:#569cd6}.hljs-link{color:#569cd6;text-decoration:underline}.hljs-built_in,.hljs-type{color:#4ec9b0}.hljs-class,.hljs-number{color:#b8d7a3}.hljs-meta .hljs-string,.hljs-string{color:#d69d85}.hljs-regexp,.hljs-template-tag{color:#9a5334}.hljs-formula,.hljs-function,.hljs-params,.hljs-subst,.hljs-title{color:#dcdcdc}.hljs-comment,.hljs-quote{color:#57a64a;font-style:italic}.hljs-doctag{color:#608b4e}.hljs-meta,.hljs-meta .hljs-keyword,.hljs-tag{color:#9b9b9b}.hljs-template-variable,.hljs-variable{color:#bd63c5}.hljs-attr,.hljs-attribute{color:#9cdcfe}.hljs-section{color:gold}.hljs-emphasis{font-style:italic}.hljs-strong{font-weight:700}.hljs-bullet,.hljs-selector-attr,.hljs-selector-class,.hljs-selector-id,.hljs-selector-pseudo,.hljs-selector-tag{color:#d7ba7d}.hljs-addition{background-color:#144212;display:inline-block;width:100%}.hljs-deletion{background-color:#600;display:inline-block;width:100%}
@@ -1,2 +0,0 @@
-!function(e){"function"==typeof define&&define.amd?define(e):e()}((function(){"use strict";var e=new Map;function t(t){var o=e.get(t);o&&o.destroy()}function o(t){var o=e.get(t);o&&o.update()}var r=null;"undefined"==typeof window?((r=function(e){return e}).destroy=function(e){return e},r.update=function(e){return e}):((r=function(t,o){return t&&Array.prototype.forEach.call(t.length?t:[t],(function(t){return function(t){if(t&&t.nodeName&&"TEXTAREA"===t.nodeName&&!e.has(t)){var o,r=null,n=window.getComputedStyle(t),i=(o=t.value,function(){s({testForHeightReduction:""===o||!t.value.startsWith(o),restoreTextAlign:null}),o=t.value}),l=function(o){t.removeEventListener("autosize:destroy",l),t.removeEventListener("autosize:update",a),t.removeEventListener("input",i),window.removeEventListener("resize",a),Object.keys(o).forEach((function(e){return t.style[e]=o[e]})),e.delete(t)}.bind(t,{height:t.style.height,resize:t.style.resize,textAlign:t.style.textAlign,overflowY:t.style.overflowY,overflowX:t.style.overflowX,wordWrap:t.style.wordWrap});t.addEventListener("autosize:destroy",l),t.addEventListener("autosize:update",a),t.addEventListener("input",i),window.addEventListener("resize",a),t.style.overflowX="hidden",t.style.wordWrap="break-word",e.set(t,{destroy:l,update:a}),a()}function s(e){var o,i,l=e.restoreTextAlign,a=void 0===l?null:l,d=e.testForHeightReduction,u=void 0===d||d,c=n.overflowY;if(0!==t.scrollHeight&&("vertical"===n.resize?t.style.resize="none":"both"===n.resize&&(t.style.resize="horizontal"),u&&(o=function(e){for(var t=[];e&&e.parentNode&&e.parentNode instanceof Element;)e.parentNode.scrollTop&&t.push([e.parentNode,e.parentNode.scrollTop]),e=e.parentNode;return function(){return t.forEach((function(e){var t=e[0],o=e[1];t.style.scrollBehavior="auto",t.scrollTop=o,t.style.scrollBehavior=null}))}}(t),t.style.height=""),i="content-box"===n.boxSizing?t.scrollHeight-(parseFloat(n.paddingTop)+parseFloat(n.paddingBottom)):t.scrollHeight+parseFloat(n.borderTopWidth)+parseFloat(n.borderBottomWidth),"none"!==n.maxHeight&&i>parseFloat(n.maxHeight)?("hidden"===n.overflowY&&(t.style.overflow="scroll"),i=parseFloat(n.maxHeight)):"hidden"!==n.overflowY&&(t.style.overflow="hidden"),t.style.height=i+"px",a&&(t.style.textAlign=a),o&&o(),r!==i&&(t.dispatchEvent(new Event("autosize:resized",{bubbles:!0})),r=i),c!==n.overflow&&!a)){var f=n.textAlign;"hidden"===n.overflow&&(t.style.textAlign="start"===f?"end":"start"),s({restoreTextAlign:f,testForHeightReduction:!0})}}function a(){s({testForHeightReduction:!0,restoreTextAlign:null})}}(t)})),t}).destroy=function(e){return e&&Array.prototype.forEach.call(e.length?e:[e],t),e},r.update=function(e){return e&&Array.prototype.forEach.call(e.length?e:[e],o),e});var n=r;document.addEventListener("alpine:init",(()=>{var e;(e=window.Alpine).directive("autosize",((e,{modifiers:t},{cleanup:o})=>{n(e);const r=Array.from(e.attributes);let i=!1;for(let{nodeName:e}of r)if("wire:model"===e||e.startsWith("wire:model.")){i=!0;break}!e.hasAttribute("wire:ignore")&&i&&e.setAttribute("wire:ignore","");const l=()=>{n.update(e)};e.addEventListener("autosize",l),o((()=>{n.destroy(e),e.removeEventListener("autosize",l)}))})),e.magic("autosize",(e=>t=>{(t||e).dispatchEvent(new Event("autosize"))}))}))}));
-//# sourceMappingURL=alpine-autosize.min.js.map
--- a/Show More
+++ b/Show More
				`@@ -1 +0,0 @@`
				(()=>{function o(e){e.directive("intersect",e.skipDuringClone((t,{value:i,expression:l,modifiers:n},{evaluateLater:r,cleanup:c})=>{let s=r(l),a={rootMargin:x(n),threshold:f(n)},u=new IntersectionObserver(d=>{d.forEach(h=>{h.isIntersecting!==(i==="leave")&&(s(),n.includes("once")&&u.disconnect())})},a);u.observe(t),c(()=>{u.disconnect()})}))}function f(e){if(e.includes("full"))return .99;if(e.includes("half"))return .5;if(!e.includes("threshold"))return 0;let t=e[e.indexOf("threshold")+1];return t==="100"?1:t==="0"?0:Number(`.${t}`)}function p(e){let t=e.match(/^(-?[0-9]+)(px\|%)?$/);return t?t[1]+(t[2]\|\|"px"):void 0}function x(e){let t="margin",i="0px 0px 0px 0px",l=e.indexOf(t);if(l===-1)return i;let n=[];for(let r=1;r<5;r++)n.push(p(e[l+r]\|\|""));return n=n.filter(r=>r!==void 0),n.length?n.join(" ").trim():i}document.addEventListener("alpine:init",()=>{window.Alpine.plugin(o)});})();
				`@@ -1 +0,0 @@`
				pre code.hljs{display:block;overflow-x:auto;padding:1em}code.hljs{padding:3px 5px}.hljs{background:#1e1e1e;color:#dcdcdc}.hljs-keyword,.hljs-literal,.hljs-name,.hljs-symbol{color:#569cd6}.hljs-link{color:#569cd6;text-decoration:underline}.hljs-built_in,.hljs-type{color:#4ec9b0}.hljs-class,.hljs-number{color:#b8d7a3}.hljs-meta .hljs-string,.hljs-string{color:#d69d85}.hljs-regexp,.hljs-template-tag{color:#9a5334}.hljs-formula,.hljs-function,.hljs-params,.hljs-subst,.hljs-title{color:#dcdcdc}.hljs-comment,.hljs-quote{color:#57a64a;font-style:italic}.hljs-doctag{color:#608b4e}.hljs-meta,.hljs-meta .hljs-keyword,.hljs-tag{color:#9b9b9b}.hljs-template-variable,.hljs-variable{color:#bd63c5}.hljs-attr,.hljs-attribute{color:#9cdcfe}.hljs-section{color:gold}.hljs-emphasis{font-style:italic}.hljs-strong{font-weight:700}.hljs-bullet,.hljs-selector-attr,.hljs-selector-class,.hljs-selector-id,.hljs-selector-pseudo,.hljs-selector-tag{color:#d7ba7d}.hljs-addition{background-color:#144212;display:inline-block;width:100%}.hljs-deletion{background-color:#600;display:inline-block;width:100%}