Commit Graph

  • 14b4e51a7c Improved quantized matrix vector product (#786) Angelos Katharopoulos 2024-03-05 17:32:19 -08:00
  • cbcf44a4ca Some fixes in cache / thread safety (#777) Awni Hannun 2024-03-05 13:30:50 -08:00
  • 859ae15a54 Fix test (#785) Awni Hannun 2024-03-04 23:02:27 -08:00
  • 0787724c44 Fast Inference SDPA op (#735) Brian Keene 2024-03-05 00:06:11 -05:00
  • 7b463ffb07 Ios compile (#784) Awni Hannun 2024-03-04 20:02:26 -08:00
  • 6686e61ca4 Reduce update (#783) Jagrit Digani 2024-03-04 19:09:51 -08:00
  • c096a77b9b revision bump (#778) v0.5.1 Awni Hannun 2024-03-04 13:41:53 -08:00
  • 5121f028d9 nice tensordot for mlx c (#782) Awni Hannun 2024-03-04 09:51:02 -08:00
  • 6a665ea6ed Dilation for convolutional layers (#766) Piotr Rybiec 2024-03-04 15:43:00 +01:00
  • bc06cb9ff6 Pickle + dtype fix for numpy conversion (#763) Awni Hannun 2024-03-02 06:09:29 -08:00
  • 8e281c76c3 Fix the top-k op (#768) Angelos Katharopoulos 2024-03-01 22:08:43 -08:00
  • d5964a2710 bindings for memory info (#761) Awni Hannun 2024-03-01 19:51:58 -08:00
  • cf3eb87e52 Fix typo in transforms.cpp (#764) Ikko Eltociear Ashimine 2024-03-01 15:23:46 +09:00
  • ab3a466711 bump (#760) v0.5.0 Awni Hannun 2024-02-29 11:58:54 -08:00
  • 4494970f47 avoid nested closures in module (#759) Awni Hannun 2024-02-29 09:39:52 -08:00
  • 776c3d226d Convolution update (#651) Jagrit Digani 2024-02-28 20:11:16 -08:00
  • f5f18b704f fix temporary bug (#752) Awni Hannun 2024-02-27 17:44:39 -08:00
  • 420ff2f331 Add back compiled function signatures and docstrings (#749) Awni Hannun 2024-02-27 13:18:59 -08:00
  • 56ba3ec40e fix cpu compile on older OS (#747) Awni Hannun 2024-02-26 22:20:53 -08:00
  • de3d2467a3 Update: Fast GeLU Approximation (#744) Noah Kasmanoff 2024-02-27 00:08:50 -05:00
  • fe1dabf272 Fix compile with non standard types (#745) Awni Hannun 2024-02-26 19:28:53 -08:00
  • 08226ab491 added atleast *args input support (#710) Hinrik Snær Guðmundsson 2024-02-26 14:17:59 -05:00
  • 3b661b7394 Add linear warmup and schedule joining for use with existing schedules (#721) Chime Ogbuji 2024-02-26 10:28:48 -05:00
  • e6418781ab Fix logsumexp edge case (#740) Awni Hannun 2024-02-25 08:39:55 -08:00
  • ac02cf33bd Fix some issues using MLX in C++ (#739) Awni Hannun 2024-02-24 22:20:57 -08:00
  • 22364c40b7 Upsample2d (#414) Gabrijel Boduljak 2024-02-23 18:55:04 +01:00
  • d729a1991b Fix arange with inf step (#686) Noah Farr 2024-02-23 15:18:15 +01:00
  • 126c9869c8 Implement the 'where' primitive for conditional selection (#664) Rifur13 2024-02-22 18:10:48 -05:00
  • ad4a45e615 Fix the release builds in CI (#729) Angelos Katharopoulos 2024-02-22 14:09:13 -08:00
  • 04fc896016 version bump (#727) v0.4.0 Awni Hannun 2024-02-22 11:54:17 -08:00
  • 884b4ed43b Fix threadgroup memory in arg reduce (#723) Jagrit Digani 2024-02-21 19:42:16 -08:00
  • 972d9a3aea Up to 10x faster scatter. (#709) Vijay Krish 2024-02-21 11:09:30 -08:00
  • 7dcdd88e27 Change the logo and add a dark option (#716) Angelos Katharopoulos 2024-02-20 10:57:02 -08:00
  • 8120a3b65c link to other APIs (#715) Awni Hannun 2024-02-20 09:54:49 -08:00
  • 5798256fcf Shapeless compilation for some graphs (#687) Awni Hannun 2024-02-19 21:43:54 -08:00
  • d0fda82595 fix tolist for half types (#702) Awni Hannun 2024-02-19 09:44:27 -08:00
  • f883fcede0 Added support for atleast_1d, atleast_2d, atleast_3d (#694) Hinrik Snær Guðmundsson 2024-02-19 12:40:52 -05:00
  • e1bdf6a8d9 discover doctests in cmake (#703) Diogo 2024-02-19 10:03:56 -05:00
  • 1a4f4c5ea6 Refactor CPU compile preamble (#708) Awni Hannun 2024-02-19 06:12:53 -08:00
  • 0925af43b0 Remove unused variables (#706) Jack Mousseau 2024-02-18 12:50:10 -08:00
  • dc937b8ed3 CPU compile (#691) Awni Hannun 2024-02-17 06:54:32 -08:00
  • c3965fc5ee Separate fast ops and primitives (#699) Awni Hannun 2024-02-16 19:16:39 -08:00
  • bf7cd29970 version bump (#698) v0.3.0 Awni Hannun 2024-02-16 08:44:08 -08:00
  • a000d2288c feat: update black pre-commit hook to 24.2.0 (#696) Nripesh Niketan 2024-02-16 18:01:59 +04:00
  • 165abf0e4c Auto-run PRs from contributors (#692) Mike Drob 2024-02-15 19:30:35 -06:00
  • 818cda16bc Support LR schedulers (#334) Srimukh Sripada 2024-02-15 20:26:20 +01:00
  • 85143fecdd improved error msg for invalid axis(mx.split) (#685) toji 2024-02-15 20:55:38 +05:30
  • 35431a4ac8 Adds device context manager (#679) Diogo 2024-02-14 17:14:58 -05:00
  • ccf1645995 Custom primitive + RoPE fat op (#676) Awni Hannun 2024-02-14 14:04:25 -08:00
  • 1a48713d32 Update gather and scatter to not use Argument Encoder (#683) Jagrit Digani 2024-02-14 13:42:13 -08:00
  • 1eb04aa23f Fix empty array construction in cpp (#684) Awni Hannun 2024-02-13 23:34:17 -08:00
  • 0c65517e91 Return empty array when repeats is 0 in mx.repeat (#681) Noah Farr 2024-02-14 02:49:31 +01:00
  • 2fdc2462c3 Faster gather and scatter. (#682) Vijay Krish 2024-02-13 17:47:41 -08:00
  • be6e9d6a9f Fixed wording in extensions.rst (#678) Hinrik Snær Guðmundsson 2024-02-13 11:39:02 -05:00
  • e54cbb7ba6 Pooling layers (#357) Gabrijel Boduljak 2024-02-13 07:08:13 +01:00
  • 40c108766b Quantized matmul fix (#677) Angelos Katharopoulos 2024-02-12 18:54:21 -08:00
  • 4cc70290f7 PR Builder Workflow (#659) Mike Drob 2024-02-12 19:47:21 -06:00
  • 74caa68d02 nit in readme (#675) Awni Hannun 2024-02-12 12:25:04 -08:00
  • 3756381358 Faster bfloat quantized mat-vec and vec-mat (#663) Awni Hannun 2024-02-11 21:53:16 -08:00
  • d12573daa6 quote file name (#670) Awni Hannun 2024-02-11 10:33:30 -08:00
  • 0dbc4c7547 feat: Update pre-commit-config.yaml (#667) Nripesh Niketan 2024-02-11 18:08:20 +04:00
  • 06072601ce Scatter optimization : Eliminate 64b integer divide. (#662) Vijay Krish 2024-02-10 08:49:51 -08:00
  • 11d2c8f7a1 Linux build for CI of other packages (#660) Angelos Katharopoulos 2024-02-09 18:17:04 -08:00
  • 7f3f8d8f8d Fix the softmax fix (#661) Awni Hannun 2024-02-09 17:02:13 -08:00
  • b96be943dc bug fix (#658) Awni Hannun 2024-02-09 16:50:45 -08:00
  • b670485185 Remainder negative numerator bug fixed (#641) Abdussamet Türker 2024-02-10 03:49:14 +03:00
  • b57bd0488d Metadata support for safetensors (#639) Diogo 2024-02-08 22:33:15 -05:00
  • 221f8d3fc2 Bump the version to 0.2 (#656) v0.2.0 Angelos Katharopoulos 2024-02-08 11:27:12 -08:00
  • 5c03efaf29 Compile docs (#653) Awni Hannun 2024-02-08 11:21:50 -08:00
  • 7dccd42133 updated calls to use loc &scale (#643) LeonEricsson 2024-02-08 18:01:59 +01:00
  • 1b97b2958b Compile with capture (#629) Awni Hannun 2024-02-07 17:29:22 -08:00
  • e5e816a5ef fix sequential with empty modules at end (#647) Awni Hannun 2024-02-07 13:22:27 -08:00
  • 28eac18571 Kernel generation (#614) Angelos Katharopoulos 2024-02-07 13:15:59 -08:00
  • 5fd11c347d Add loc and scale to random.normal (#638) Noah Farr 2024-02-07 20:49:59 +01:00
  • ef73393a19 Feat: Add weights argument in BCE Loss and tests (#620) Aryan Gupta 2024-02-07 23:09:52 +05:30
  • ea406d5e33 CI change (#645) Angelos Katharopoulos 2024-02-07 06:04:34 -08:00
  • 146bd69470 Skip compile when transforming (#635) Awni Hannun 2024-02-05 21:28:37 -08:00
  • 316ff490b3 Remove masks from BlockLoader and clear out load case for invalid thread (#634) Jagrit Digani 2024-02-05 16:00:17 -08:00
  • d40a04f8dc minor fixes (#631) Awni Hannun 2024-02-05 13:27:49 -08:00
  • d75ae52ecd Compile primitive (#571) Awni Hannun 2024-02-05 06:51:22 -08:00
  • 31fea3758e feat: enhancement of the error message for mlx.core.mean (#608) Avikant Srivastava 2024-02-05 14:51:49 +05:30
  • e319383ef9 Faster gather (#626) Awni Hannun 2024-02-04 17:25:44 -08:00
  • 5c3ac52dd7 fix test (#627) Awni Hannun 2024-02-04 16:18:03 -08:00
  • ebfd3618b0 fixes for building and running on iOS (#619) David Koski 2024-02-04 12:29:17 -08:00
  • 11a9fd40f0 fix: handle linspace function when num is 1 (#602) Avikant Srivastava 2024-02-05 00:33:49 +05:30
  • 4fd2fb84a6 make python array SupportsAbs conform (like numpy) (#624) Daniel Strobusch 2024-02-04 18:31:02 +01:00
  • 9852af1a19 fix "shape" docstring. (#623) Daniel Strobusch 2024-02-04 18:21:22 +01:00
  • 16750f3c51 Fix typo in CMakeLists.txt (#616) minghuaw 2024-02-03 05:59:26 -08:00
  • 95b5fb8245 minor changes (#613) Awni Hannun 2024-02-02 11:48:35 -08:00
  • 83f63f2184 Add Margin Ranking Loss (#536) AtomicVar 2024-02-03 02:57:31 +08:00
  • cb6156d35d Fix eval in trace bugs (#612) Awni Hannun 2024-02-02 09:57:12 -08:00
  • 506d43035c typo fix (#607) Piotr Rybiec 2024-02-02 02:39:55 +01:00
  • 36cff34701 Bump the version (#604) v0.1.0 Angelos Katharopoulos 2024-02-01 11:41:38 -08:00
  • e88e474fd1 Reduce vmap + some fixes (#601) Awni Hannun 2024-02-01 11:30:28 -08:00
  • 601c6d6aa8 Fix for AdaDelta (#603) David Koski 2024-02-01 09:56:27 -08:00
  • ba8d6bf365 Change the transformer to norm_first by default (#599) Angelos Katharopoulos 2024-01-31 12:55:30 -08:00
  • 4a5f3b21bb Add py.typed to support PEP-561 (type-hinting) for mlx (#588) Sugato Ray 2024-01-31 15:05:42 -05:00
  • fcc5ac1c64 Add GPU support for uint64/int64 reductions (#569) Vijay Krish 2024-01-31 11:18:04 -08:00
  • bad67fec37 Added TeX line breaks to mlx.optimizers.Lion docstring (#595) nathan 2024-01-30 19:37:34 -08:00
  • 199aebcf77 Change the variance computation (#319) Angelos Katharopoulos 2024-01-30 19:28:56 -08:00