Commit Graph

Select branches

Hide Pull Requests

geohot/updates

gh-pages

master

remove_anchors

#1

#2

#3

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

• 4a07ea355d buffer options should work (#3211) George HotzandGitHub 2024-01-22 19:23:55 -08:00
• a06f34ae42 remove dead lines from cstyle (#3212) George HotzandGitHub 2024-01-22 18:59:19 -08:00
• 8465938d29 minor hlb_cifar cleanups (#3208) chenyuandGitHub 2024-01-22 12:38:39 -05:00
• 3378625773 name upcast variables (#3200) David HouandGitHub 2024-01-22 08:37:28 -08:00
• 827b7a3c64 cleanup pad_reflect and make_square_mask in hlb_cifar (#3206) chenyuandGitHub 2024-01-22 11:30:46 -05:00
• 99884f4c98 cifar flags for RANDOM_CROP, RANDOM_FLIP, and CUTMIX (#3204) chenyuandGitHub 2024-01-22 01:12:51 -05:00
• 53afec2841 add HALF to handcode_resnet50_opt.py (#3202) chenyuandGitHub 2024-01-21 23:03:59 -05:00
• 836883fedc comment out cutmix in hlb_cifar (#3201) chenyuandGitHub 2024-01-21 22:24:53 -05:00
• e6c71f1b26 fix device of Tensor.arange inside Tensor.one_hot (#3199) chenyuandGitHub 2024-01-21 21:03:50 -05:00
• f7d1c42239 cleanup noop prefixes in _pool (#3198) chenyuandGitHub 2024-01-21 20:03:32 -05:00
• 640e5c36ad Fix metal tests broken by 3f56d1a (#3196) uuuvnandGitHub 2024-01-21 18:53:25 +02:00
• b9d27636aa cleanup test_ops.py (#3192) chenyuandGitHub 2024-01-20 20:08:56 -05:00
• 3f56d1a5e8 add operator.lt and operator.eq to test_dtype_alu (#3191) chenyuandGitHub 2024-01-20 14:54:02 -05:00
• c4b5661146 fuzz length for multitensor reduce test case (#3190) chenyuandGitHub 2024-01-20 00:44:38 -05:00
• fdb1c2b1d9 move reduce over 0 len axis logic to lazy.py (#3188) chenyuandGitHub 2024-01-20 00:13:03 -05:00
• 485332935e ring copy example (#3185) chenyuandGitHub 2024-01-19 23:34:30 -05:00
• 254a7372fe buffer copy refactor (#3187) George HotzandGitHub 2024-01-19 20:21:24 -08:00
• fb4bd2a57d reenable padto to search action (#3183) chenyuandGitHub 2024-01-19 14:17:53 -05:00
• cb4cfc078a parameterize multitensor tests for reduce (#3181) chenyuandGitHub 2024-01-19 14:03:01 -05:00
• 5097d5b808 fix padto when with late reduce (#3180) nimlgenandGitHub 2024-01-19 22:01:44 +03:00
• 729a01bf3e complex PRs will not be merged geohot 2024-01-19 10:58:47 -08:00
• f87ecbb0f3 fuzzer validates outputs + (partially) oob accesses (#3178) nimlgenandGitHub 2024-01-19 21:34:51 +03:00
• b2571d586c hypothesis.st -> hypothesis.strat (#3179) chenyuandGitHub 2024-01-19 11:55:26 -05:00
• c4faedebf3 add test cases for negative entry max allreduce (#3177) chenyuandGitHub 2024-01-18 22:26:51 -05:00
• ab1b7c4d09 fix allreduce for max (#3175) chenyuandGitHub 2024-01-18 20:25:35 -05:00
• c51c90bcd4 more sync in transfer (#3174) George HotzandGitHub 2024-01-18 17:17:03 -08:00
• 28dcbf0e00 test case sharded batchnorm has different ast on devices (#3172) chenyuandGitHub 2024-01-18 18:12:15 -05:00
• a60d50487d disable padto, seems to have bug in gpt2 (#3173) chenyuandGitHub 2024-01-18 18:09:30 -05:00
• c80884884e event driven hip (#3160) George HotzandGitHub 2024-01-18 14:35:18 -08:00
• d2aab65958 remove unused expr node (#3170) George HotzandGitHub 2024-01-18 14:18:43 -08:00
• 097b1390ec touchup test_indexing (#3169) chenyuandGitHub 2024-01-18 14:32:43 -05:00
• a04e4d0442 inline clang renderer (#3168) George HotzandGitHub 2024-01-18 11:17:34 -08:00
• efbe4788d1 indexing: Final cleanup (#3156) geohotstanandGitHub 2024-01-19 03:08:03 +08:00
• e139ae550d smaller limit_dims_to_max (#3167) chenyuandGitHub 2024-01-18 13:02:20 -05:00
• 992067399e clean up exceptions in __del__ everywhere (#3165) nimlgenandGitHub 2024-01-18 19:34:09 +03:00
• 0338903429 Update kits19.py (#3166) Max-WeandGitHub 2024-01-18 17:33:50 +01:00
• 67bc2ddfd8 JIT cleanups (#3164) George HotzandGitHub 2024-01-17 23:39:57 -08:00
• f0c178b7e9 move get_contraction to helpers (#3162) George HotzandGitHub 2024-01-17 19:13:11 -08:00
• e52a609240 make WINO a context var, and LATEWINO in hlb_cifar (#3161) chenyuandGitHub 2024-01-17 20:21:26 -05:00
• ee83505fcc fix test extra issue (#3159) George HotzandGitHub 2024-01-17 11:58:08 -08:00
• 9cc2577a08 use hip events (#3157) George HotzandGitHub 2024-01-17 10:39:57 -08:00
• 1b508e0f71 fix fuzz_linearizer toCPU to as_buffer (#3158) chenyuandGitHub 2024-01-17 13:18:46 -05:00
• 743b36f0ce hotfix: copy size is in bytes geohot 2024-01-17 16:44:15 +00:00
• 2e6162b281 graph cleanup (#3155) George HotzandGitHub 2024-01-16 20:57:31 -08:00
• a72b1b6d65 sharding for llama (#3151) 3 people 2024-01-16 19:28:00 -08:00
• 14c010958b support for non-uniform sharding (#3154) 3 people 2024-01-16 20:33:32 -05:00
• 81ae4ea179 compile cache for several devices (#3148) nimlgenandGitHub 2024-01-16 22:45:26 +03:00
• 589c16756f hlb_cifar multi gpu training (#3150) chenyuandGitHub 2024-01-16 14:38:45 -05:00
• cc0de99751 hotfix: multilazybuffer can have only one lazybuffer geohot 2024-01-16 10:06:45 -08:00
• 228f30b96a multitensor jit (#3149) 3 people 2024-01-16 09:09:15 -08:00
• b9d470577c gelu -> quick_gelu in hlb_cifar (#3147) chenyuandGitHub 2024-01-16 02:03:37 -05:00
• ec5a212b0a modernize hlb_cifar (#3146) chenyuandGitHub 2024-01-16 01:35:11 -05:00
• 2088937206 run full hlb_cifar training in tinybox ci (#3145) chenyuandGitHub 2024-01-15 23:59:20 -05:00
• 22920a7e55 add LATEBEAM to hlb_cifar (#3142) chenyuandGitHub 2024-01-15 23:26:03 -05:00
• 766bd0bbe8 make _deepwalk a generator and not passing nodes around (#3141) chenyuandGitHub 2024-01-15 20:26:00 -05:00
• 120c8b1841 update llvm api + add cache key (#3140) George HotzandGitHub 2024-01-15 17:25:32 -08:00
• cec0a7bc37 use shard api to eval resnet fast (#3136) George HotzandGitHub 2024-01-15 16:49:38 -08:00
• ca0beeef38 Christopherm99 ptx (#3139) +212 people 2024-01-15 16:44:20 -08:00
• 1ee11411f1 s/lazydata/lazyop/ in print_tree (#3138) chenyuandGitHub 2024-01-15 19:38:27 -05:00
• a5d634a541 simplify dtype (#3137) George HotzandGitHub 2024-01-15 16:27:43 -08:00
• e4528543fa remove LLVMOPT geohot 2024-01-15 16:01:09 -08:00
• a464909d79 fast resnet eval (#3135) George HotzandGitHub 2024-01-15 14:15:18 -08:00
• b7b494e9b8 no numpy (#3134) Jyotirmaya MahantaandGitHub 2024-01-16 02:39:05 +05:30
• 6bb65cd02e fix off-by-one error in st_equal (#3131) Paul GustafsonandGitHub 2024-01-15 11:32:13 -08:00
• 44c05919c1 dtype fmt (#3132) George HotzandGitHub 2024-01-15 11:31:54 -08:00
• 5ec66938de remove np from metal graph (#3129) nimlgenandGitHub 2024-01-15 19:44:35 +03:00
• 2ef09ca641 update test_ptr_ne (#3130) Jyotirmaya MahantaandGitHub 2024-01-15 22:06:29 +05:30
• e39cd3e7f2 update env_vars.md (#3127) chenyuandGitHub 2024-01-15 01:06:56 -05:00
• 537fb8b0b8 separate try except blocks in onnx2torch in model benchmark (#3126) chenyuandGitHub 2024-01-15 00:39:33 -05:00
• 0dba34b81c Fix backward fn for < and == (#3037) 3 people 2024-01-15 04:39:52 +00:00
• db965a0c74 remove numpy from ops_torch (#3124) chenyuandGitHub 2024-01-14 22:46:57 -05:00
• 1f9aee8b6f remove numpy from device (#3123) George HotzandGitHub 2024-01-14 19:36:05 -08:00
• ea5824657d move fromcpu out of lazy.py (#3122) George HotzandGitHub 2024-01-14 18:21:08 -08:00
• 96345061d3 hotfix: ptrdtype compare was broken geohot 2024-01-14 18:08:22 -08:00
• 26e0faf656 make DType a dataclass (#3111) Jyotirmaya MahantaandGitHub 2024-01-15 06:45:59 +05:30
• c13d51da1d add device options for tests in multigpu (#3121) Yixiang GaoandGitHub 2024-01-14 15:17:47 -08:00
• 79f4627fbc fix conversation: llama generates token not prob now (#3120) chenyuandGitHub 2024-01-14 13:10:01 -05:00
• 152ef7fc79 minor cleanups of onnx_ops (#3116) chenyuandGitHub 2024-01-14 02:15:24 -05:00
• fb3f8f7597 move sample inside jit for beautiful_mnist (#3115) chenyuandGitHub 2024-01-14 01:36:30 -05:00
• a313e63a9b add Tensor.var (#3114) chenyuandGitHub 2024-01-14 01:11:08 -05:00
• c658aa4fbf minor cleanup of test_disk_tensor (#3112) chenyuandGitHub 2024-01-13 20:54:58 -05:00
• 9c73d2724f cleanup ops_disk type annotation and redundant str cast (#3110) chenyuandGitHub 2024-01-13 16:56:48 -05:00
• a300fea2a4 failed test case due to cast resets shapetracker (#3109) chenyuandGitHub 2024-01-13 12:46:51 -05:00
• cf1d0a6704 no exceptions in __del__ when module creation is failed in hip/cuda (#3107) nimlgenandGitHub 2024-01-13 20:03:55 +03:00
• 12f28ac9d4 catch runtime error in search._time_program (#3106) chenyuandGitHub 2024-01-12 21:53:13 -05:00
• f018a55ea1 update NumNode.__hash__ to be hash(self.b) (#3105) chenyuandGitHub 2024-01-12 19:46:21 -05:00
• c3c35f9142 flag to profile mixtral - 1.7 tok/s now (#3104) chenyuandGitHub 2024-01-12 18:54:27 -05:00
• e078e2d060 add half @ half to mac benchmark (#3103) chenyuandGitHub 2024-01-12 16:38:41 -05:00
• ddbdb52f77 wmma: enable METAL half tensor cores and clean up cstyle (#3095) Francis LamandGitHub 2024-01-12 13:25:28 -08:00
• f96fc6e9d4 fix gpt2 with empty prompt take 2 (#3102) chenyuandGitHub 2024-01-12 14:46:36 -05:00
• ca46d3541b Revert "fix gpt2 with empty prompt" (#3101) chenyuandGitHub 2024-01-12 14:27:41 -05:00
• 1d7f01bc6d fix gpt2 with empty prompt (#3100) chenyuandGitHub 2024-01-12 14:18:17 -05:00
• 0c49d38ba7 replace with tensor op (#3099) SnakeOnexandGitHub 2024-01-12 20:13:40 +01:00
• f3a50b4e40 fix broadcasted logic if there's 0 in shapes (#3097) chenyuandGitHub 2024-01-12 13:32:43 -05:00
• 025fbf4e80 One hot in tensor.py (#3093) SnakeOnexandGitHub 2024-01-12 19:31:18 +01:00
• 7086d77db1 bugfix do not reset shapetracker of 0 size lazybuffer (#3096) chenyuandGitHub 2024-01-11 23:22:52 -05:00
• 13e872b53f add mutigpu support for llama attention (#3064) Yixiang GaoandGitHub 2024-01-11 16:31:02 -08:00
• dcf7ecaaff update jit type annotation post lazy rewrite (#3091) chenyuandGitHub 2024-01-11 15:49:30 -05:00
• 0fe6904351 use device from LinearizerOptions in kernel search (#3090) chenyuandGitHub 2024-01-11 14:46:03 -05:00
• 93e3f952aa use BEAM=2 instead of BEAM=4 in cuda ci gpt2 (#3089) chenyuandGitHub 2024-01-11 13:21:06 -05:00