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

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

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

delta_attention

delta_attention_scan

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

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

gl_dims

gldims

glm_flash

grad_outerworld

grok_x86_novec

group_div_rules

hcq2_vram_copies

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_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

no_weakint_variable

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

outer_range

outerworld_work

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

qwen_mergable

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_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

vf_ish

view_after_const

view_in_codegen

viz_cbroadcast

warp_fun

wgpu-f16-packed

working_postopt

x86_moves

x86_numel

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

• f0cc2b66f8 add an aneccompile example in Objective-C (#240) freedom" Koan-Sin TanandGitHub 2021-02-01 00:31:16 +08:00
• eabe0b9017 remove deepwalk args (#243) Göktuğ KarakaşlıandGitHub 2021-01-31 19:30:17 +03:00
• ce77dda805 yolov5 v4 geohot 2021-01-05 07:56:17 -08:00
• 62e3a8558c fix tolerance maybe geohot 2021-01-05 07:45:47 -08:00
• 1c148f2fe4 fixed example broken after gpu refactor (#238) AsimandGitHub 2021-01-05 20:41:54 +05:00
• 8a38e0d207 only mish failed geohot 2021-01-03 09:47:11 -08:00
• a337f7780e smarter way to write sign geohot 2021-01-03 09:46:00 -08:00
• 1a4487965a remove negative from things w/o negative geohot 2021-01-03 09:43:34 -08:00
• 0531b848eb second class sign geohot 2021-01-03 09:33:12 -08:00
• 0702e0c763 nah, no sign, it's not what you want. use relu geohot 2021-01-03 09:30:33 -08:00
• 29655609d5 fix GPU sign...these tests aren't very good geohot 2021-01-03 09:00:49 -08:00
• ea9c9af5d7 faster sign geohot 2021-01-03 08:54:21 -08:00
• c2eeb6950b add support for sign. technically relu can be second class now geohot 2021-01-03 08:29:57 -08:00
• 6842ad9ec8 minor cleanups, yolo work geohot 2021-01-03 08:14:16 -08:00
• 0825cf7f79 ⚡ Added softplus and mish non stable (#220) NeuralLinkandGitHub 2021-01-03 21:38:41 +05:30
• ac229ea750 remove print geohot 2021-01-02 12:53:30 -08:00
• 895d142503 start trying to load yolo v5 geohot 2021-01-02 12:51:45 -08:00
• ece07a3d12 🔨 refactor register ops (#233) NeuralLinkandGitHub 2021-01-02 21:17:16 +05:30
• 42b4761025 transformer >99.98% test accuracy in ~30s (#230) Marcel BischoffandGitHub 2021-01-02 10:45:09 -05:00
• ebd72ff437 Test split (#231) LiamandGitHub 2021-01-01 15:19:03 +01:00
• 4a7cf2e420 more reordering geohot 2020-12-31 09:58:02 -05:00
• 92abe43683 reduce before binary because of unbroadcasting geohot 2020-12-31 09:49:52 -05:00
• 4291002881 reorder GPU ops geohot 2020-12-31 09:46:39 -05:00
• de7fe085de no read out of bounds geohot 2020-12-31 09:41:36 -05:00
• 1fb5fcafce GPU slice should fix tests geohot 2020-12-31 09:37:03 -05:00
• e972a45456 Dynamically register ops to Tensor (#232) LiamandGitHub 2020-12-31 15:10:19 +01:00
• e2f833f58f max to behave on ties like torch (#229) Marcel BischoffandGitHub 2020-12-30 18:52:50 -05:00
• 30f8132646 reorder ops in ops cpu geohot 2020-12-30 11:00:01 -05:00
• e5b2803b5d ops in readme geohot 2020-12-30 10:48:55 -05:00
• 2d44bf7f1a Dot -> Matmul geohot 2020-12-30 10:41:51 -05:00
• 10fc3ff5b9 cleaner syntax geohot 2020-12-30 10:35:37 -05:00
• fcfe3dae01 write slice for CPU geohot 2020-12-30 10:32:53 -05:00
• 47504004fd ane ops geohot 2020-12-29 18:00:53 -05:00
• 1f5c9618ef refactor in readme and issue #225 geohot 2020-12-29 17:30:04 -05:00
• f9170505b3 if you like your transformers twice as slow, use the GPU geohot 2020-12-29 17:14:23 -05:00
• 6a6a82e999 support multidot on GPU geohot 2020-12-29 16:56:30 -05:00
• 27208d729b add GPU max thanks to marcelbischoff geohot 2020-12-29 16:44:14 -05:00
• 4bbad11afe link to papers geohot 2020-12-29 14:15:46 -05:00
• 3f8e137b6f extra/transformer geohot 2020-12-29 14:14:00 -05:00
• c4e7a1ae59 accessors are dumb geohot 2020-12-29 14:10:26 -05:00
• fb6aaefb9b save 2 lines geohot 2020-12-29 14:02:50 -05:00
• ea341c84fe logsoftmax good, div bad geohot 2020-12-29 13:59:39 -05:00
• f18801c7db simple pool. swimming is very easy now geohot 2020-12-29 13:48:50 -05:00
• 8f9232d59b readmee geohot 2020-12-29 13:40:34 -05:00
• 837aaacfbf Unpad2D on GPU: geohot 2020-12-29 13:16:14 -05:00
• 02655c07d5 break maxpool2d on GPU geohot 2020-12-29 13:05:57 -05:00
• 061e37de39 touchups geohot 2020-12-29 12:41:21 -05:00
• a2e6562330 fix max op, less lines geohot 2020-12-29 10:47:04 -05:00
• dc8fa7999c Transpose on GPU (#221) Marcel BischoffandGitHub 2020-12-29 10:40:11 -05:00
• 36579f66bf max op geohot 2020-12-28 23:54:52 -05:00
• bcb3ceeca3 set training in functions geohot 2020-12-28 22:45:46 -05:00
• 51bf164b72 dropout, training geohot 2020-12-28 22:12:23 -05:00
• 7b8fee038d it works! forgot the sqrt geohot 2020-12-28 16:23:52 -05:00
• 1faf05ef67 ahh, it's better if i don't train the embedding geohot 2020-12-28 16:07:02 -05:00
• c3832e1bde hmm, fix layernorm to not be batchnorm and it breaks geohot 2020-12-28 13:06:21 -05:00
• 2e89e75dcb layernorm fixes transformer instability geohot 2020-12-28 12:58:15 -05:00
• 628d21f899 doc touchup geohot 2020-12-28 10:45:26 -05:00
• fafece9db7 avgpool2d is a second class op geohot 2020-12-28 10:41:59 -05:00
• 593233b668 log and exp are first class ops geohot 2020-12-28 10:00:30 -05:00
• ffff98db78 Evaluation in Transformers (#218) Marcel BischoffandGitHub 2020-12-28 09:24:51 -05:00
• 65b07d2f4f fix onehot embed geohot 2020-12-27 18:50:38 -05:00
• d864e1c71a transformer is training geohot 2020-12-27 18:46:32 -05:00
• a361ef6861 fixup training loop geohot 2020-12-27 18:35:56 -05:00
• f15bec6dbc make multidot work on CPU geohot 2020-12-27 17:25:37 -05:00
• 131e04c90c cpu only decorator geohot 2020-12-27 17:18:55 -05:00
• 2f1b2c0a3b add transpose, start on transformer geohot 2020-12-27 16:59:12 -05:00
• d379502c04 Cleaning (#211) 3 people 2020-12-27 15:58:51 +01:00
• 8a335f03ad clock speed 32x32 geohot 2020-12-22 18:18:52 -05:00
• aae2e35208 benchmarking 512x512 GEMM geohot 2020-12-22 18:01:36 -05:00
• bd18e03138 conv from weights works geohot 2020-12-22 17:42:17 -05:00
• b3cf53e39b more docs geohot 2020-12-22 17:14:38 -05:00
• 4065eae0fb docs for tensor stride geohot 2020-12-22 17:06:36 -05:00
• 6fb127d5c7 l2 cache note geohot 2020-12-22 16:48:19 -05:00
• 78a06a1285 more readme geohot 2020-12-22 16:23:08 -05:00
• 0ab951f21c better readme geohot 2020-12-22 15:57:33 -05:00
• 6ca449afd2 sum works geohot 2020-12-22 12:53:20 -05:00
• ebc7f8305c 3x3 gemm in conv geohot 2020-12-22 12:00:44 -05:00
• ff52f832dc conv hwx geohot 2020-12-22 11:49:54 -05:00
• 5b93d9c03a aneregs geohot 2020-12-22 10:41:12 -05:00
• ccfd03e401 parse ane struct geohot 2020-12-22 10:16:46 -05:00
• 06f359baa3 issue-193 - Move torch loader out of efficientnet code (#213) Nicklas BomanandGitHub 2020-12-22 06:19:16 +01:00
• 3abda29648 add debug to ane lib geohot 2020-12-21 18:49:33 -05:00
• c48917fdf0 ane reg notes geohot 2020-12-21 15:26:52 -05:00
• 451e944648 aneregs from debug geohot 2020-12-21 13:37:02 -05:00
• 16f0cb7d80 fix sign python geohot 2020-12-20 15:44:10 -05:00
• e03759f01d channels concat geohot 2020-12-19 09:43:21 -08:00
• 73ae6d7821 working on gemm geohot 2020-12-19 09:30:29 -08:00
• 075cf4bc02 Update examples to new API (#205) LiamandGitHub 2020-12-19 18:04:12 +01:00
• 3c9150a9f4 lil note geohot 2020-12-18 09:54:20 -08:00
• be41e73ad9 ane docs continue geohot 2020-12-18 09:49:13 -08:00
• e5f2055776 ane refactor, fix examples geohot 2020-12-18 09:19:06 -08:00
• fbcd1912cf ane struct geohot 2020-12-18 09:06:25 -08:00
• 56d44637f3 fixed pylint, formatted python files iwth cblack on localhost (#204) 3 people 2020-12-17 17:37:31 -05:00
• 799ad5ba17 refactor/softmax (#201) 3 people 2020-12-17 12:53:28 -05:00
• dad061dafb Added Python 3 style super() without arguments (#200) 3 people 2020-12-17 05:50:16 +01:00
• bcf1518309 All devices are equal! (#196) LiamandGitHub 2020-12-16 08:44:08 +01:00
• 78210b5e40 less lines (#197) James RobertsandGitHub 2020-12-14 23:53:00 +02:00
• b86bbd2e72 readmes geohot 2020-12-13 21:32:20 -08:00
• da72a0eed4 Big MNIST model with PIL augmentation and load/save (#160) Marcel BischoffandGitHub 2020-12-13 23:45:55 -05:00
• f50dcc12ac 1k lines geohot 2020-12-13 20:37:58 -08:00