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

• 5bbcbd145c Add cumsum with n-dim inputs (#922) Tom EdwardsandGitHub 2023-06-05 00:55:23 +01:00
• 091bd65a68 feat: quick doc fixups (#923) wozeparrotandGitHub 2023-06-04 14:03:57 -04:00
• fbf17f0031 intel benchmark matmul gets 60 TFLOPS? geohot 2023-06-04 17:01:50 +00:00
• 657e642e3a Fixed test suite for Clip (#912) Steven AndersonandGitHub 2023-06-04 09:01:01 -07:00
• 0b88c5f923 Eliminate LoadOps.FROMCPU (#920) kposborne2andGitHub 2023-06-04 08:55:50 -07:00
• 3e0b37f050 randn slow geohot 2023-06-04 08:52:13 -07:00
• e9c1ae3825 Add a quick start guide (#900) wozeparrotandGitHub 2023-06-04 11:51:20 -04:00
• d429553730 Allow Tensor(tuple) (#911) Alexey ZaytsevandGitHub 2023-06-04 13:48:19 +07:00
• afd0be8a9c intel example geohot 2023-06-04 06:43:09 +00:00
• 2b4baa97e9 exclude string type from external_test_onnx_backend.py (#918) MohammedAlkhrashiandGitHub 2023-06-04 05:10:52 +03:00
• b78addf2f8 Whisper (#919) George HotzandGitHub 2023-06-03 18:55:14 -07:00
• ed1963b899 Fast DiskTensor to other Tensor (#916) George HotzandGitHub 2023-06-03 12:25:41 -07:00
• 791530045d Refactor LoadOps (#910) George HotzandGitHub 2023-06-03 09:40:43 -07:00
• d58586bb17 safetensors! (#903) George HotzandGitHub 2023-06-02 13:41:09 -07:00
• 513aeb2f66 Fixed all ConstantOfShape test suite (#907) Steven AndersonandGitHub 2023-06-02 11:26:40 -07:00
• 301f7b54c6 ConstantOfShape ONNX test fixed. (#890) Steven AndersonandGitHub 2023-06-02 07:34:25 -07:00
• 5feee9c94b Fix .std() tests on torch=1.13 (#904) Alexey ZaytsevandGitHub 2023-06-02 21:33:51 +07:00
• 4d28d55683 add nn layer tests geohot 2023-06-01 21:34:24 -07:00
• 8a928ed2f3 nn init matches torch (#901) George HotzandGitHub 2023-06-01 21:24:11 -07:00
• 27845fd3a3 train_efficientnet: only import datasets.imagenet when IMAGENET is set (#899) Peter RossandGitHub 2023-06-02 12:19:52 +10:00
• 1b42b4e1b8 fix examples/hlb_cifar10.py geohot 2023-06-01 19:03:17 -07:00
• bfea5215e9 Add weight decay to SGD (#883) wozeparrotandGitHub 2023-06-01 16:13:18 -04:00
• 0e9e0fd718 document environment variables (#887) Nicklas BomanandGitHub 2023-06-01 22:11:17 +02:00
• 67a7674787 added conv1d tests -> simple, padding, stride, asymmetric padding (#896) SnakeOnexandGitHub 2023-06-01 22:10:37 +02:00
• ea9d4e6d78 Slightly improved readability of reshape method condition check (#898) Bartłomiej JargutandGitHub 2023-06-01 22:10:08 +02:00
• dd41f3ee40 compact the global dimensions using the shapetracker (#897) George HotzandGitHub 2023-06-01 13:09:54 -07:00
• ef129bcb85 Zero dim Tensor support (#777) JoqsanandGitHub 2023-06-01 21:32:02 +03:00
• ae83e9844c add output_padding to transposed conv (#875) kposborne2andGitHub 2023-06-01 00:03:22 -07:00
• 740304ef9d Small Onnx Parser Improvements (#885) Friedrich Carl EichenrothandGitHub 2023-06-01 09:01:01 +02:00
• 115903a15c Add unbiased std and corresponding tests (#881) Tom EdwardsandGitHub 2023-06-01 00:32:36 +01:00
• 447b5847e2 Added test for empty tensor for Tensor.numel(), added missing numel call (#880) 3 people 2023-05-31 21:28:47 +02:00
• b58d875937 Add Tensor.ndim .element_size .is_floating_point (#876) Alexey ZaytsevandGitHub 2023-05-31 23:00:35 +07:00
• 3924aae8ed Fix ONNX dropout and unify the implementation (#857) Marcello FuschiandGitHub 2023-05-31 16:40:47 +02:00
• 1272d8526a Llvm int support (#866) DiogoandGitHub 2023-05-30 20:49:26 -04:00
• 5670123d88 Add tensor.numel (#869) Nima KhodaveisiandGitHub 2023-05-31 01:08:09 +02:00
• 2e393f7ef2 InstanceNormalization ONNX test fixed. (#870) skobsmanandGitHub 2023-05-31 00:07:44 +01:00
• 0dab8edc97 support Int64 type in cstyle gen (#860) DiogoandGitHub 2023-05-30 19:04:46 -04:00
• 0fc4cf72a2 feat: add train scaffolding (#859) wozeparrotandGitHub 2023-05-30 10:10:40 -04:00
• f91f28d9e2 fix a bunch of tests (#856) Friedrich Carl EichenrothandGitHub 2023-05-30 02:48:26 +02:00
• 502e33652f add Tensor.full and Tensor.full_like and reuse them (#852) Ubaidullah KhanandGitHub 2023-05-30 02:48:09 +02:00
• dd1ac5d267 stack based context manager for DEBUG and IMAGE (#844) 3 people 2023-05-30 02:36:15 +02:00
• 3075988468 Added kaiming_uniform initialization for Conv2d and Linear layers (#756) Rabia Eda YılmazandGitHub 2023-05-30 01:09:55 +03:00
• 174c65b7d9 add onnx Binarizer op (#850) 3 people 2023-05-29 16:15:50 -04:00
• 4408c25e9a Add Onnx op Shrink (#851) M4tthewDEandGitHub 2023-05-29 22:15:39 +02:00
• 6f2b3755ca set axis default to 0 (#854) Friedrich Carl EichenrothandGitHub 2023-05-29 22:15:28 +02:00
• 3b158f7a5f fix onnx versions greater or equal 10 (#853) Friedrich Carl EichenrothandGitHub 2023-05-29 22:04:06 +02:00
• 0e89c3f456 zeros_like use dtype if specified else default to tensor’s dtype (#848) Ubaidullah KhanandGitHub 2023-05-29 20:38:34 +02:00
• 1a5d72f812 Onnx ops And, Or, Xor, Not (#847) DiogoandGitHub 2023-05-29 14:09:20 -04:00
• 844e6d0753 conv1d & conv3d onnx tests (#835) SnakeOnexandGitHub 2023-05-29 19:16:45 +02:00
• ae204e40c8 move line counter to python geohot 2023-05-29 09:21:18 -07:00
• 8c6085a715 Rewrite Adam/W as functions of LAMB (#839) wozeparrotandGitHub 2023-05-29 12:21:35 -04:00
• ddc9dafe62 tighten up the kernel count tests geohot 2023-05-29 08:48:54 -07:00
• 9a18dade4b Fix ResNet Kernel Fusion (from bounty) (#825) vaibhavandGitHub 2023-05-29 21:16:36 +05:30
• f3168ee69b default transformer dropout to 0 (#828) JudeDavis1andGitHub 2023-05-29 16:06:16 +01:00
• c825cc4774 use tensor dtype for zeros_like() (#842) Ubaidullah KhanandGitHub 2023-05-29 17:05:50 +02:00
• 7925fa58d9 Fix cuda (#836) crthilakrajandGitHub 2023-05-29 16:59:36 +02:00
• 6ea5df19b2 Fix conv_transpose2d asymmetric padding (#840) Marcello FuschiandGitHub 2023-05-29 16:57:06 +02:00
• 2fd2fb6380 int8/uint8 support (#837) wozeparrotandGitHub 2023-05-29 02:15:06 -04:00
• 2939e40b98 Count Python tokens (#817) Ali BenkassouandGitHub 2023-05-29 06:26:49 +02:00
• 5d212864b5 Add MLPerf UNet3D model (#775) Jacky LeeandGitHub 2023-05-28 20:38:19 -07:00
• 65d09031f2 add retinanet with resnet backbone (#813) SohaibandGitHub 2023-05-29 03:20:16 +00:00
• 46327f7420 bugfix for stable diffusion geohot 2023-05-29 00:03:09 +00:00
• 59f9bcd4a4 Disktensors! (#819) George HotzandGitHub 2023-05-28 15:40:37 -07:00
• 6d49925a26 Add max_pool2d dilation (#833) Marcello FuschiandGitHub 2023-05-29 00:16:48 +02:00
• 7460bd9b02 Add LAMB optimizer (#821) wozeparrotandGitHub 2023-05-28 18:09:05 -04:00
• 1b337b5533 ONNX tests exclude all unsupported filetype tests (#832) SnakeOnexandGitHub 2023-05-28 22:31:20 +02:00
• 21d27d31a9 Fix a couple pad tests (#827) cheeetooandGitHub 2023-05-28 13:06:46 -06:00
• 081b3ab639 Tensor.where method (#830) Kirill RandGitHub 2023-05-28 20:20:33 +03:00
• eea3542975 remove other install method geohot 2023-05-28 08:36:21 -07:00
• 0c0c7380af Add Tensor.where (#826) Kirill RandGitHub 2023-05-28 18:04:56 +03:00
• 2163a1b049 Add shrink step to fix strided conv_transpose2d, and add to nn (#823) 3 people 2023-05-28 07:52:45 -07:00
• 01daa74f9b fixed TestCustomFunction (#820) crthilakrajandGitHub 2023-05-28 03:27:46 +02:00
• 67de3aa1de Add mlperf bert model (#803) wozeparrotandGitHub 2023-05-27 17:53:32 -04:00
• 1e56aced05 add changeable DEBUG (#816) George HotzandGitHub 2023-05-27 13:28:25 -07:00
• a3feee29c5 make tests faster + add onnx (#815) George HotzandGitHub 2023-05-27 08:53:32 -07:00
• 606b841d3f LR Schedulers (#755) Mattis MegevandandGitHub 2023-05-27 16:47:49 +02:00
• 87fa5af70a ptx example George Hotz 2023-05-09 13:11:56 -07:00
• fd296ce444 have kernels wait on DEBUG=1 geohot 2023-05-26 22:51:16 +00:00
• 8b2c2d6896 Optimizations in symbolic.py (#796) Rayan HatoutandGitHub 2023-05-26 20:59:53 +01:00
• 26014a0fa1 add convtranspose (#809) George HotzandGitHub 2023-05-26 12:35:03 -07:00
• 04284414db Batchnorm2d fixed running_var (#807) symlonandGitHub 2023-05-26 21:32:49 +02:00
• 65d63f5b40 support folding multiple of 4 into float4 (#808) George HotzandGitHub 2023-05-26 12:17:48 -07:00
• 6d4a728f62 Don't collapse dimensions during batched matmul (FIX #799) (#800) Aneesh DurgandGitHub 2023-05-26 13:15:34 -05:00
• 803587b8b4 update readme geohot 2023-05-26 06:10:41 +00:00
• 7351eb4b61 feat: put temperary file in the same directory as the destination file (#805) wozeparrotandGitHub 2023-05-25 23:46:02 -04:00
• 3ddcb5c36f Half4 load (#804) George HotzandGitHub 2023-05-25 20:21:15 -07:00
• ee2c8423c7 disable that test on LLVM. i have to stop pushing to master geohot 2023-05-26 03:11:03 +00:00
• ea3194f68e test touchups geohot 2023-05-26 02:39:42 +00:00
• 0dc333cfab Promote Embedding to nn (#798) wozeparrotandGitHub 2023-05-25 21:39:45 -04:00
• f4f23dc9a3 version bump v0.6.0 geohot 2023-05-26 00:51:25 +00:00
• faf80418b7 pyopencl by default since GPU is default (#802) George HotzandGitHub 2023-05-25 17:48:18 -07:00
• fca5028d78 feat: ability to exclude cl devices from being used (#801) wozeparrotandGitHub 2023-05-25 20:31:29 -04:00
• 3c465470f2 pip installation one liner (#793) BenediktandGitHub 2023-05-26 01:43:42 +02:00
• a968c4c3a4 Cleanup mlperf (#797) George HotzandGitHub 2023-05-25 11:36:43 -07:00
• c19ef0fcce Add sin/cos/tan (#794) DiogoandGitHub 2023-05-25 12:04:56 -04:00
• 01ae45a43c Add mlperf RNN-T model (#782) wozeparrotandGitHub 2023-05-25 03:41:21 -04:00
• b258af117a Fix PytestCollectionWarning when running tests (#791) Sasha KrassovskyandGitHub 2023-05-24 22:17:57 -08:00
• 0400315078 Revert "ops rdna" geohot 2023-05-21 13:02:18 -07:00
• 325a3bf2cf Revert "writing 2" geohot 2023-05-21 13:02:17 -07:00
• dddd6c42f0 writing 2 geohot 2023-05-21 12:52:36 -07:00