Commit Graph

  • 67e319488c fix python tests Awni Hannun 2023-12-26 14:47:56 -0800
  • 49e3e99da3 remove unused includes Awni Hannun 2023-12-26 10:58:02 -0800
  • f7cea9563d some style and API consistency updates to linalg norm Awni Hannun 2023-12-26 10:54:59 -0800
  • 4bae4a8239 removed unused imports Gabrijel Boduljak 2023-12-24 04:40:44 +0100
  • bbfe042a2b fixed a bug with no ord and axis provided Gabrijel Boduljak 2023-12-22 12:00:18 +0100
  • 5a184d5b5d more tests Gabrijel Boduljak 2023-12-22 05:34:06 +0100
  • f82ab0eec9 reused existing util for implementation of linalg.norm Gabrijel Boduljak 2023-12-22 04:58:51 +0100
  • 145a4d143d refactored mlx.linalg.norm bindings Gabrijel Boduljak 2023-12-22 04:35:02 +0100
  • 2831c77b33 added better docs and examples Gabrijel Boduljak 2023-12-22 02:39:47 +0100
  • 49c48de53b handling inf, -inf as numpy does, more extensive tests of compatibility with numpy Gabrijel Boduljak 2023-12-22 01:19:57 +0100
  • 26bb16e768 added some tests for python bindings Gabrijel Boduljak 2023-12-22 00:33:36 +0100
  • fa096d64a2 updated python bindings Gabrijel Boduljak 2023-12-21 19:09:36 +0100
  • b996d682d9 removed unused import in linalg.cpp Gabrijel Boduljak 2023-12-21 18:36:50 +0100
  • 5d7a06717c added tests Gabrijel Boduljak 2023-12-21 18:34:02 +0100
  • 8c43d820d9 completed the implementation of the norm Gabrijel Boduljak 2023-12-21 18:33:23 +0100
  • 05203ecd78 renamed vector_norm to norm, implemented norm without provided ord Gabrijel Boduljak 2023-12-20 03:13:18 +0100
  • 24da85025f implemented vector_norm python binding Gabrijel Boduljak 2023-12-17 07:06:04 +0100
  • cc9b2dc3c2 implemented vector_norm in cpp Gabrijel Boduljak 2023-12-17 02:55:33 +0100
  • 447bc089b9
    Fix tolerance in de-/quantization test (#295) Angelos Katharopoulos 2023-12-26 19:21:05 -0800
  • f0bf2bf09a Fix tolerance in de-/quantization test Angelos Katharopoulos 2023-12-26 17:59:33 -0800
  • 47a64c480b
    Merge branch 'main' into transformer_fix YUN, Junwoo 2023-12-27 02:18:34 +0800
  • 0372cbbcbf precommit junwoo-yun 2023-12-27 02:16:26 +0800
  • bee11e79d9 add doctstring junwoo-yun 2023-12-27 02:15:24 +0800
  • 31e563dbc9
    Merge pull request #1 from ml-explore/main David Joffe 2023-12-26 19:20:17 +0200
  • fc4e5b476b
    Fix llama link in README.md (#289) Yutaka Kondo 2023-12-25 20:53:20 -0800
  • 49c9226d99
    Fix llama link in README.md Yutaka Kondo 2023-12-26 13:48:59 +0900
  • d58ac083f3
    expose itemsize and nbytes as for numpy arrays (#284) Daniel Strobusch 2023-12-25 19:34:28 +0100
  • 6824099cfb
    expose itemsize and nbytes as for numpy arrays Daniel Strobusch 2023-12-25 17:59:36 +0100
  • a123c3c7d2
    implement-batch-norm-layer (#217) __mo_san__ 2023-12-25 16:32:53 +0100
  • 6b4f49fe1c cleanup stats test Awni Hannun 2023-12-25 07:27:14 -0800
  • 865e53fcab doc nits Awni Hannun 2023-12-25 07:07:24 -0800
  • 88a94b9db8 run precommit junwoo-yun 2023-12-25 07:52:19 +0800
  • 0e0557b756 run precommit junwoo-yun 2023-12-25 07:50:54 +0800
  • 297e69017c add docstring, activation, norm_first junwoo-yun 2023-12-25 07:39:42 +0800
  • 9e6b8c9f48
    Refactor the reduction kernels (#277) Angelos Katharopoulos 2023-12-24 14:47:57 -0800
  • 15577cb727
    Update __init__.py __mo_san__ 2023-12-24 23:14:04 +0100
  • a1c06b7d46 updated BN implementation to handle input shape as NLC and NWHC^^ m0saan 2023-12-24 23:05:18 +0100
  • 9bf68814a4 updated BN implementation to handle input shape as NLC and NWHC^^ m0saan 2023-12-24 23:04:31 +0100
  • 28009c9cdb Update python/mlx/nn/layers/normalization.py __mo_san__ 2023-12-24 22:22:26 +0100
  • cf5a5a4a01 updated the batch norm doc string ^^ m0saan 2023-12-23 23:08:10 +0100
  • c68a472b83 Update python/mlx/nn/layers/normalization.py __mo_san__ 2023-12-23 15:20:08 +0100
  • 019a85511c improve batch norm code ^^ m0saan 2023-12-22 20:58:54 +0100
  • b444a6a693 Update normalization.py __mo_san__ 2023-12-22 20:57:32 +0100
  • a43b853194 refactored and updated batch norm tests ^^ m0saan 2023-12-22 20:50:05 +0100
  • 8b08f440d9 Update __init__.py __mo_san__ 2023-12-22 10:02:29 +0100
  • 02ce72d4cd Update layers.rst __mo_san__ 2023-12-22 10:02:00 +0100
  • 82ca771e69 updated BN implementation to be more generic ^^ m0saan 2023-12-22 09:58:04 +0100
  • 7b0f8bda9c updated docs and added examples to doc string ^^ m0saan 2023-12-19 10:13:24 +0100
  • 7ec3cadf98 added test cases for batch norm on 3D input & refactored code ^^ m0saan 2023-12-19 10:06:35 +0100
  • eca773b62c Update normalization.py __mo_san__ 2023-12-19 09:17:50 +0100
  • a0b2a34e98 rebasing ... m0saan 2023-12-19 09:16:53 +0100
  • d4bf9a2976 calc running mean and var only when training m0saan 2023-12-19 06:29:52 +0100
  • c3c2fcf41d update batch norm implementation -> fixed some bug and added support for 3D inputs m0saan 2023-12-19 06:27:14 +0100
  • e9fd1cf02d update batch norm implementation m0saan 2023-12-19 04:41:34 +0100
  • ad53687ae7 Update normalization.py __mo_san__ 2023-12-18 23:49:52 +0100
  • 2b617b63bd implemented batchnorm layer m0saan 2023-12-18 23:26:37 +0100
  • 22fee5a383
    Remove redundant assert in losses.py (#281) Zach Schillaci 2023-12-24 11:39:08 -0500
  • 185403538a Remove redundant assert in losses.py Zachary Schillaci 2023-12-24 11:28:03 -0500
  • 7365d142a3
    random.uniform must respect dtype, even if lower precision than "low" (#280) Daniel Strobusch 2023-12-24 16:04:43 +0100
  • 0ea828dff6
    random.uniform must respect dtype, even if lower precision than "low" Daniel Strobusch 2023-12-24 15:17:48 +0100
  • 93ea457a08 fix styling Bahaa Eddin tabbakha 2023-12-24 11:42:02 +0300
  • c9328d5c68 add repeat function Bahaa Eddin tabbakha 2023-12-24 11:37:15 +0300
  • 3f710299b5 Refactor the reduction kernels Angelos Katharopoulos 2023-12-23 14:12:31 -0800
  • 8b227fa9af
    fix no metal build (#276) Awni Hannun 2023-12-23 19:18:10 -0800
  • ace3a0ac5a fix no metal build Awni Hannun 2023-12-23 16:37:10 -0800
  • 8c3da54c7d
    Fix failing test for log cosh loss (#275) Vidit Agarwal 2023-12-24 05:56:46 +0530
  • 6c3c16e07b reformatted by pre-commit black vidit 2023-12-24 04:53:02 +0530
  • 4d3c451b3f fix assert statement in log_cosh_loss vidit 2023-12-24 04:38:03 +0530
  • acf1721b98
    Corrected the example of value_and_grad (#274) Vidit Agarwal 2023-12-24 00:36:38 +0530
  • 5952ebcb92 Reformat through pre-commit/black vidit 2023-12-24 00:32:41 +0530
  • 9ca34c6287 Corrected the example for mx.value_and_grad vidit 2023-12-24 00:08:17 +0530
  • 51322b90e3 comment Awni Hannun 2023-12-23 06:42:41 -0800
  • 1baff8c2b4 lighter weight json Awni Hannun 2023-12-23 06:39:55 -0800
  • 680f18cca5 switch statement Awni Hannun 2023-12-22 21:15:11 -0800
  • 313f6bd9b1 change name to safetensors Awni Hannun 2023-12-22 21:06:49 -0800
  • f91f450141
    Fix argmax returns documentation (#263) Finn Voorhees 2023-12-22 23:33:17 -0500
  • dc2cee1572
    Fix argmax returns documentation Finn Voorhees 2023-12-22 20:58:35 -0500
  • 227ef82784 idk why i removed this dc-dc-dc 2023-12-22 20:20:18 -0500
  • 172e696540 remove message from test output Awni Hannun 2023-12-22 17:03:48 -0800
  • c6d7702ef0 remove pathlib refs dc-dc-dc 2023-12-22 19:06:31 -0500
  • b9cbfbf3d7 Added docstring for scale parameter Hazem 2023-12-23 00:22:44 +0200
  • 0e5ae1d1bf Added RoPE scaling test Hazem 2023-12-23 00:20:28 +0200
  • a415bac4f7 Ran pre-commit Hazem 2023-12-23 00:06:59 +0200
  • 273d558117 Added scale for RoPE Hazem 2023-12-23 00:05:55 +0200
  • ee6ce00aee docs and made retain_graph optional bool dc-dc-dc 2023-12-22 16:19:31 -0500
  • fa093967ec moved to subdirectory dc-dc-dc 2023-12-22 16:05:07 -0500
  • cd3616a463
    Revisit autorelease memory pools (#260) Ronan Collobert 2023-12-22 11:01:26 -0800
  • d9478d0eb0 new_memory_pool -> new_scoped_memory_pool Ronan Collobert 2023-12-22 10:51:49 -0800
  • d35fa1db41
    Add Hinge, Huber and LogCosh losses (#199) Nicholas Santavas 2023-12-22 19:28:10 +0100
  • 5ccc1fb314 no metal backend support Ronan Collobert 2023-12-22 09:44:05 -0800
  • 29a8b2047b make things simpler Ronan Collobert 2023-12-21 16:22:52 -0800
  • 433de52bf3
    review nsantavas 2023-12-22 18:24:08 +0100
  • a813bdda0a make general autorelease pool part of metal device Ronan Collobert 2023-12-18 20:45:29 -0800
  • e8deca84e0
    Add dropout2d (#250) Justin Deschenaux 2023-12-22 17:02:29 +0100
  • 9bc8930084 Add dropout2d Justin Deschenaux 2023-12-21 23:28:29 +0100
  • a8152c1a25
    review nsantavas 2023-12-21 09:43:25 +0100
  • 1733e17b57
    Add Hinge, Huber and LogCosh losses nsantavas 2023-12-17 20:38:52 +0100
  • 13fd7bd721 small fixes dc-dc-dc 2023-12-22 01:25:42 -0500
  • fdf9d99f0f add back retain_graph argument dc-dc-dc 2023-12-20 14:39:31 -0500
  • edd55388ce switch python load test to use load dc-dc-dc 2023-12-20 13:56:55 -0500