Commit Graph

Select branches

Hide Pull Requests

abstractions_4

add_local_buffer

after_qol

algebraic_upat

all_bv_no_idx

all_sink

amd_isel

amd_refac

amd_smu_reset_patch

amd_sqtt

amd_uop

amd_uop_cleanups

assign_copy

assign_try_3

autogen-comgr-2

before2

better_spec

better_syn

bitcast_spec

broadcast_shape_2

bt_ei

buf_heur

bump_nv_2404

cache_reshape

call_inline

call_is_kernels

callify

cast_vec_as_cast

cat_mop

changes_from_rangify

check_clsetarg

check_dims

ci-deps

cifar_30

cifar_30_clean

clean_load

clean_spec

clone_tg

codegen_try_2

contig_on_after

continue_mat

deepsek

default_late_allreduce

delete_kernel_py

delete_ones

delete_slow_rangeify

demote_op

disable_rules

dsp_search

dsp_search_merged

dtype_shape

earlier_gater

emu_cdna

emulate_ctx

end_all_ranges

endrange

er_prioity

execution_unit

explicit-dev

failing_assign

fancy_usb

fast_llm

fix_fa_fwd_clang_22

fix_fuse

fix_llvm_cast

fix_range_merging_sd

fix_rangeify_tests

fix_rc

fix_sched_copy

fix_slice_store

fix_src_spec

fixup_fix_kernel

full_spec

fun_w_egraphs

gate-on-load-store

gate_rewrite

gemma4_gpt

gen_pdf_fast

gh-pages

gitea-test-ci

gl_dims

gldims

glm_flash

grad_outerworld

grok_x86_novec

group_div_rules

heur_4096

heur_group

hk_gemm

http_threading

hwtest_fixes

id_store

if_range_cleanups

ignore_oob

image_idx_xy

image_in_mc

image_no_vec

image_refactors_stack

index_to_slice

invalid_try_3

jinja_template

kernel4

kernel_is_call

kernelless

kimi_slop

kitten_matmul

late_Locals

late_add_load

late_allreduce

late_codegen

late_decomp_is_late

late_lower_idx

late_view

late_weakint

lds_double_buffering

less_wmma_arg

lil_dg_change

lil_rangeify_td

llama_trainer

llm_262k_minimal

llm_flash_attn

llm_qwen

llm_serve_fixes

load_view_late

local_cache

locals_have_device

loop_end_op

loop_is_rng

lowerer_hacks

mac_pytest

manual_kernel_3

master

mega_lowerer

min_new_lin

min_outer_range

minor_views

mixins7

mlperf_training_v4.1

mlperf_training_v5.0

mlperf_training_v6.0

modern_scan

more_apps

more_early_comps

more_folding_rules

more_kernelcount

more_pcode

more_rangeify_pm_tests

more_ren_clean

more_uop_gc_test

more_uop_mm

more_vmap

move_devec

move_dtype_decomps

move_expander

move_gates_to_load_store

move_mop_cleanup

move_schedule

move_simp_views

moveleftright

mt_typing

multi_range

multioutput

new_ck

new_devec

new_invalid_and_image

new_linearizer

new_shape

new_test

no-chat-template

no_assign_2

no_buf_0

no_decimals

no_late_simplify_marg

no_merge_ends

no_merge_views

no_mstack_in_allreduce

no_one_ranges

no_sip_usbgpu

no_tsink_base

no_tuplize

no_user_ck_contig

no_vec_x86_4

noop_index

novecimg

num_batches_tracked_shape

olmo3_support

one_compiler

one_hcopt

only_assign_buffers

only_loop_backends

only_reg_emu

only_reg_emu2

onnx-cpu-count

outer_range

outerworld_work

pad_invalid

pad_invalid_work

pad_mixin

param_call_is_function

parse_sqtt_more

pci_iommu

port_nir

postrange_hcopts

precompile_backward

prefill_qwen_fast

print_special_ops

ps_clean

pyremu_try

python_prg

python_speed

q4k_support

qcom-openpilot-13ms

qcom_830

qcom_mmapeak

qcom_speed_codex

qol_string

qwen36_27b_amd_900

qwen36_27b_amd_minimal

qwen36_amd_bugfixes

r_cleanups

r_new_opt

ranged_call

rangeify_2

rangeify_const_folding

rdna3_vibes

rdna4_asm

rdna4_gemm

real_pyrender

recurse_sub

reg_coalese

relu_pattern

remove_assign

remove_cast_hack

remove_ck_contig

remove_define_var

remove_forced_re

remove_on_stack

remove_programspec

remove_pylint

remove_slow_tests

remove_vec_2

remove_vec_3

renderers_use_shapes_2

replace_if_with_range

reprocess_node_experiments

reshape_trait

resnet-ci

revert-11547-fix_mismatch_red

revert-11715-revert-11711-amd_llvm_tc_cast

revert-11809-no_ki_in_gpudims

revert-11811-better_div_nesting

revert-11876-memplan_vbig

revert-12104-simplify_valid_in_symbolic

revert-12308-bufferize_early

revert-12731-jitlink_init_check

revert-12813-after_cleanups

revert-13046-better_cleanup_arange_buffers

revert-13311-prof_link_sqtt

revert-13583-rocr_desc_2

revert-14734-tload

revert-17141-full_rewrite_to_sink

revert-17408-_broadcasted-weak

revert-9438-simple_matmul_np_rand

revert_wrong_ck_fix

rewrite_rangeify2

rtoposort

scan_assign

scan_op_8

sched_cache

sched_lin

sd_load_simple

self_type

set_testpath

shrink_in_render

simp_spec

simple_gidxs

simple_pool

simple_priority

simple_rangeify

simpler-sd-compile

simpler_end

simpler_fusion

simpler_image

simpler_postrange

simpler_reduce

slice_like_shrink

spec_3

spec_cleanup

spec_ren

speed_one_linear

split_buf_idx

split_decomp_pass

split_sub

stack_0

store_is_void

string_ranges

support_opts_in_contig

swizzle_in_kernel

sym_llm

sym_work

symbolic-chunk-split

szdiff_bot_test

tc_is_heur

tensor_graph_single_sink

test-lp

test_amd_crash

test_fa

test_from_postopt

test_hgemm

test_unit

tiny_scan

tmr

tool_call_support

topoprop

ttomsa-new_linearizer

tuplegtuple

two_stage_remove

typed_checks

typing_work

unused_permute_arg

uop_prg

upd_test4

update_benchmark

update_benchmark_search

update_mlperf

user_failure

vecless_x86_glm

verify_sink_dims_clean

view_after_const

view_in_codegen

viz_cbroadcast

warp_fun

wgpu-f16-packed

working_postopt

x86_moves

x86_numel

#1

#2

#3

#4

#5

v0.10.0

v0.10.1

v0.10.2

v0.10.3

v0.11.0

v0.12.0

v0.13.0

v0.4.0

v0.5.0

v0.6.0

v0.7.0

v0.8.0

v0.9.0

v0.9.1

v0.9.2

Mono Color

• 8ce8664024 Golfed -1 (#315) namlizandGitHub 2022-04-05 20:16:27 +03:00
• a6203b5ecd Correct spelling mistakes (#314) Edward BettsandGitHub 2022-04-05 13:22:18 +01:00
• 050636bcb1 here's two extra lines of precious code (#307) Andrei AugustinandGitHub 2022-02-21 19:45:19 +00:00
• d065040df0 Fix yolo param name (#306) Ildar KashaevandGitHub 2022-02-10 13:40:24 +05:00
• cfb7a4c41a back to 1000 lines George Hotz 2022-01-17 10:44:39 -08:00
• a482d56229 don't save for backward pass if we don't need to. add purple edges to graph geohot 2022-01-16 12:18:41 -08:00
• 8dfdc4e11f GRAPH and DEBUG can be independent. under 1k lines geohot 2022-01-16 12:04:10 -08:00
• 2f531e35be add ctx saved tensors to graph geohot 2022-01-16 11:18:36 -08:00
• 2a10116bfa support drawing graphs geohot 2022-01-16 10:45:58 -08:00
• 6a5cb6842e make profiler work for backward pass too geohot 2022-01-16 09:35:10 -08:00
• 2cae2dfa07 don't crash the dataloader for imagenet George Hotz 2022-01-16 08:41:26 -08:00
• 907ff7dbb6 imagenet is training George Hotz 2022-01-15 23:27:31 -08:00
• d1e082e0ef add imagenet training support George Hotz 2022-01-15 23:16:38 -08:00
• ade2af7ac0 data doesn't require grad George Hotz 2022-01-15 22:41:27 -08:00
• 0973e54eb0 fix DEBUG for GPU George Hotz 2022-01-15 22:14:28 -08:00
• d844ecee27 track running stats by default, and detach George Hotz 2022-01-15 21:46:54 -08:00
• d541e2a8e5 move assert, remove comment George Hotz 2022-01-15 21:36:58 -08:00
• 845bb1fc34 bs 4 -> 2 in training test George Hotz 2022-01-15 21:34:21 -08:00
• c0d1254003 don't run unneeded grads George Hotz 2022-01-15 21:32:13 -08:00
• ecc0903451 Revert "faster deepwalk" George Hotz 2022-01-15 21:02:18 -08:00
• 931500a098 faster deepwalk George Hotz 2022-01-15 20:57:57 -08:00
• 7025c9bbeb Transfer learning for ResNet (#295) cjg91andGitHub 2022-01-15 20:22:10 -08:00
• 55d792b065 Revert "fixup resnet" George Hotz 2022-01-15 20:22:01 -08:00
• 4eabe677ed fixup resnet George Hotz 2022-01-15 20:21:02 -08:00
• e0bef0bd01 training is False by default George Hotz 2022-01-15 19:57:41 -08:00
• 8ba3d1f803 fix bn test, affine is True George Hotz 2022-01-15 19:52:15 -08:00
• 8ec2341cca fix bn training George Hotz 2022-01-15 19:47:01 -08:00
• 0e6832a8ea support torch GPU, only autoinit cuda in the buffer George Hotz 2022-01-15 19:15:12 -08:00
• d8d19ed468 wikimedia wasn't returning 200 George Hotz 2022-01-15 19:09:29 -08:00
• 52918fbf78 cuda stub George Hotz 2022-01-15 19:02:14 -08:00
• 81664baf64 Fix OpenCL installation (#301) Jacky LeeandGitHub 2022-01-06 07:35:48 -08:00
• b0511f9392 Revert "does pybind fix CI?" geohot 2021-12-30 14:13:58 -05:00
• d128e4fcae does pybind fix CI? geohot 2021-12-30 14:11:39 -05:00
• 5efb6653c4 a bit of work on metal geohot 2021-12-30 13:53:08 -05:00
• c0c2c0b041 support larger ViT models geohot 2021-12-12 10:45:10 -08:00
• 785fe8ead7 unify torch Slice geohot 2021-11-30 16:25:37 -05:00
• e59381d0da cleanups, remove np geohot 2021-11-30 16:22:00 -05:00
• e28cdfb0cf clean up resnet geohot 2021-11-30 16:14:54 -05:00
• 8f5779eeaa very minor change geohot 2021-11-30 15:54:03 -05:00
• d31ef0ae48 make vit names match pytorch geohot 2021-11-30 11:34:14 -05:00
• 4b7c31b5b7 break vit into it's own file geohot 2021-11-30 11:19:22 -05:00
• 46bbbcf7f0 model touchups geohot 2021-11-30 11:13:34 -05:00
• 7d7e2b690d clean up Conv2d geohot 2021-11-30 11:02:55 -05:00
• 835869974c clean up vit code geohot 2021-11-30 10:58:03 -05:00
• 6884add850 save two lines geohot 2021-11-30 01:16:48 -05:00
• 9b538629bb cosmetic geohot 2021-11-30 01:01:39 -05:00
• 38dccb3a2e same simpler sum and max for gpu geohot 2021-11-30 00:59:05 -05:00
• 5d60df2b10 simpler sum and max geohot 2021-11-30 00:53:27 -05:00
• c39824bc62 oops, forgot some stars geohot 2021-11-30 00:46:14 -05:00
• 908db3bdea support bias in conv like linear geohot 2021-11-30 00:44:59 -05:00
• bd21304e3c linear takes in weight and bias geohot 2021-11-30 00:38:47 -05:00
• 535f02cc64 use sequential geohot 2021-11-30 00:25:39 -05:00
• de938c2d9d vit is now tested geohot 2021-11-30 00:23:06 -05:00
• aff810e722 unify transformer block geohot 2021-11-29 18:58:15 -05:00
• 58ed46963e fix broadcastdot geohot 2021-11-29 18:54:57 -05:00
• 033b04494a resnet pretrained is broken geohot 2021-11-29 18:13:52 -05:00
• 125e74293f promote layernorm to tensor op geohot 2021-11-29 18:08:21 -05:00
• dca076dbf1 remove dumb nn ops geohot 2021-11-29 18:05:31 -05:00
• 33720e733f support keepdim geohot 2021-11-29 17:47:43 -05:00
• 8a02bd56a1 refactor: canonicalize axis geohot 2021-11-29 17:29:18 -05:00
• 70544e7e9f sum hook override geohot 2021-11-29 17:14:24 -05:00
• 8097b8f7d6 vit works geohot 2021-11-29 16:28:14 -05:00
• 7c07c5efdd plz fix vit geohot 2021-11-29 15:45:19 -05:00
• ca160504e1 affine is always the last dim geohot 2021-11-29 15:22:49 -05:00
• e86f7a4aa3 deterministic geohot 2021-11-29 15:10:15 -05:00
• f909ab194f gelu with broken test geohot 2021-11-29 15:00:50 -05:00
• 9ce881f88c fix bug in getitem, drop int axis geohot 2021-11-29 14:01:24 -05:00
• c752033283 fix GPU OOM in test geohot 2021-11-29 13:05:59 -05:00
• 1eafa5580e layernorm with learnable parameters geohot 2021-11-29 13:03:57 -05:00
• c7f795ca1e added dot affine geohot 2021-11-29 12:55:56 -05:00
• 30eb3afbe1 add bias term to transformer geohot 2021-11-29 12:45:27 -05:00
• 99b6051467 add ff_dim to transformer geohot 2021-11-29 12:40:52 -05:00
• b4839eb6bb factor out binary_broadcast geohot 2021-11-29 03:49:12 -05:00
• ca154f8b59 add sub and a few refactors geohot 2021-11-29 03:35:23 -05:00
• 7b3c442f3b sum in metal geohot 2021-11-29 03:13:41 -05:00
• c6fb087695 fix old list behavior geohot 2021-11-29 01:49:21 -05:00
• 7c160c6cee metal: got relu working and add somewhat working geohot 2021-11-29 01:40:18 -05:00
• 29dee59368 cat: forward only not required geohot 2021-11-29 00:14:56 -05:00
• 06838481e8 use get data dtype geohot 2021-11-28 23:28:37 -05:00
• 3cdc77f526 add cat support geohot 2021-11-28 23:21:49 -05:00
• ce3d198bb7 less lines and fix default device geohot 2021-11-27 11:18:49 -05:00
• 7ae14179d3 refactor ops geohot 2021-11-27 11:12:23 -05:00
• 4320c45c4b oops, gpu not available geohot 2021-11-17 12:54:33 -08:00
• e2a8961a18 less lines, fix bug geohot 2021-11-17 12:52:17 -08:00
• e8ca3ad053 add Buffer abstraction for each device geohot 2021-11-17 12:46:07 -08:00
• c162e748f5 fix float64 warning on training geohot 2021-10-30 20:07:31 -07:00
• b0f14b4af8 move datasets into datasets geohot 2021-10-30 19:55:50 -07:00
• a95ef16c8c sub 1000 lines geohot 2021-10-30 19:48:24 -07:00
• 844540a5ed yolo in readme geohot 2021-10-30 19:47:34 -07:00
• ba28761894 move yolo into examples/yolo geohot 2021-10-30 19:46:00 -07:00
• 121d5a17ee use tinynn for Conv2d geohot 2021-10-30 19:40:44 -07:00
• 6bee5bdb7d add torch tests geohot 2021-10-30 18:58:45 -07:00
• 75260c6e49 unify more of cpu and torch geohot 2021-10-30 18:57:24 -07:00
• b1dae764d3 more similar between torch and numpy geohot 2021-10-30 18:46:59 -07:00
• d6cb373c79 import the torch ops that are the same geohot 2021-10-30 18:22:44 -07:00
• 51a49de299 tests pass for torch ops geohot 2021-10-30 18:00:12 -07:00
• 641b1dbb40 remove ane, start supporting ops_torch geohot 2021-10-30 17:47:00 -07:00
• 7d12482d80 refactor efficientnet loading geohot 2021-10-30 17:02:17 -07:00
• 7472a7ebe2 not forcing 3.9 for a stupid type geohot 2021-10-30 16:52:40 -07:00
• 114f6ca3fd more readme cleanup geohot 2021-10-30 16:51:25 -07:00