Commit Graph

  • 6ea6b4258d Fix style check (#395) Angelos Katharopoulos 2024-01-07 05:54:58 -08:00
  • 48f6ca8c3a Add theta cache for Rope and mask cache for ALiBi (#375) Anchen 2024-01-07 00:22:58 -08:00
  • c6d2878c1a safely divide for 0 size inputs (#388) Awni Hannun 2024-01-07 00:19:54 -08:00
  • b34bf5d52b fix saving for non-contiguous arrays (#389) Awni Hannun 2024-01-06 12:44:02 -08:00
  • 608bd43604 Move the matmul type check in the op (#384) Angelos Katharopoulos 2024-01-05 19:10:13 -08:00
  • 4c48f6460d Fix segfault from buffer protocol and tests (#383) Angelos Katharopoulos 2024-01-05 18:17:44 -08:00
  • 1331fa19f6 Make array conform to the Python Buffer Protocol (#323) Daniel Strobusch 2024-01-06 00:58:33 +01:00
  • dfdb284e16 make behaviour of dtype arguments consistent and compliant to numpy (#379) Daniel Strobusch 2024-01-05 18:37:46 +01:00
  • d8f41a5c0f support python mlx.array creation from list of mlx.array's (#325) mutexuan 2024-01-05 10:53:33 +08:00
  • b9e415d19c bump pre commit and fix format (#373) Awni Hannun 2024-01-04 16:28:52 -08:00
  • c82a8cc526 move all ObjC (via metal-cpp) interaction until post static initializers (#370) davidkoski 2024-01-04 16:12:00 -08:00
  • 75dc537e44 Fix the sigmoid module (#371) Angelos Katharopoulos 2024-01-04 13:16:36 -08:00
  • cf88db44b5 revert copy (#366) Awni Hannun 2024-01-04 10:43:29 -08:00
  • 16856a0160 Remove useless pass (#364) Chunyang Wen 2024-01-04 22:34:01 +08:00
  • d752f8e142 Fix CI (#359) Awni Hannun 2024-01-04 06:33:08 -08:00
  • d2467c320d Added support for python copy (#335) toji 2024-01-04 10:29:40 +05:30
  • 0d31128a44 use union instead of | (#358) Diogo 2024-01-03 22:33:19 -05:00
  • 1ac18eac20 simple numpy helper for tests (#352) Diogo 2024-01-03 22:19:19 -05:00
  • 526466dd09 version bump (#355) v0.0.7 Awni Hannun 2024-01-03 14:48:24 -08:00
  • e7f5059fe4 Support for quantized matmul with w and w^T (#349) Angelos Katharopoulos 2024-01-03 14:22:36 -08:00
  • d7ac050f4b feat: Add contributors graph to README (#332) Nripesh Niketan 2024-01-04 01:03:11 +04:00
  • c7edafb729 implemented InstanceNorm (#244) Gabrijel Boduljak 2024-01-03 21:21:15 +01:00
  • dff4a3833f Module checks the weight on load_weights (#337) Awni Hannun 2024-01-02 18:55:42 -08:00
  • 0782a4573a Add Tensordot op (#344) Diogo 2024-01-02 20:15:00 -05:00
  • af66a09bde Adds issue template with common questions (#345) Diogo 2024-01-02 19:52:20 -05:00
  • 436bec9fd9 Fix the implementation of the Bilinear layer (#347) Angelos Katharopoulos 2024-01-02 16:46:18 -08:00
  • 99c80a2c8b Memory allocation (#292) Awni Hannun 2024-01-02 11:59:19 -08:00
  • 295ce9db09 Feature expand nn linear (#315) Asaf Zorea 2024-01-02 16:08:53 +02:00
  • 44c1ce5e6a Spelling (#342) Josh Soref 2024-01-02 00:08:17 -05:00
  • 144ecff849 Remove useless import (#340) Chunyang Wen 2024-01-02 11:25:49 +08:00
  • 350095ce6e fix type cast error in item() for bfloat16 (#339) mutexuan 2024-01-02 11:02:04 +08:00
  • e09bf35b28 feat: Add Dropout3d layer to nn.layers (#313) Nripesh Niketan 2024-01-01 02:01:21 +04:00
  • 99c20f523e fix typos (#327) Daniel Strobusch 2023-12-31 15:06:47 +01:00
  • e3b8da2a49 Added implementation for Scaled RoPE. (#261) Hazem Essam 2023-12-31 16:06:01 +02:00
  • a020a2d49d Improve repeat using broadcasting and reshape (#318) Angelos Katharopoulos 2023-12-29 21:40:20 -08:00
  • 930b159885 Fix: typo in README.md (#316) Nripesh Niketan 2023-12-30 00:58:00 +04:00
  • 5ad8fb7268 feat: add softsign, softmax, hardswish, logsoftmax activation function (#309) Nripesh Niketan 2023-12-29 23:49:36 +04:00
  • 2aedf3e791 Minor refactor for tree_map and tree_unflatten (#311) Chunyang Wen 2023-12-29 12:55:10 +08:00
  • 473b6b43b4 Use defaultdict (#307) Chunyang Wen 2023-12-29 06:46:13 +08:00
  • d29770eeaa Update batchnorm to have the running stats in parameters (#305) Angelos Katharopoulos 2023-12-28 14:31:10 -08:00
  • 040c3bafab Add missing f str (#306) Chunyang Wen 2023-12-28 22:09:34 +08:00
  • 05767b026f Add information for dropout probability (#304) Chunyang Wen 2023-12-28 13:51:30 +08:00
  • a83d5d60bd Addition in acknowledgements (#302) Diogo 2023-12-27 16:46:47 -05:00
  • ff2b58e299 Add support for repeat (#278) Bahaa 2023-12-28 00:11:38 +03:00
  • 4417e37ede Transformer fix (#167) YUN, Junwoo 2023-12-28 00:48:36 +08:00
  • 79c95b6919 Fix load compilation (#298) Angelos Katharopoulos 2023-12-27 06:20:45 -08:00
  • 1f6ab6a556 Safetensor support (#215) Diogo 2023-12-27 05:06:55 -05:00
  • 6b0d30bb85 linalg.norm (#187) Gabrijel Boduljak 2023-12-27 04:42:04 +01:00
  • 447bc089b9 Fix tolerance in de-/quantization test (#295) Angelos Katharopoulos 2023-12-26 19:21:05 -08:00
  • fc4e5b476b Fix llama link in README.md (#289) Yutaka Kondo 2023-12-25 20:53:20 -08:00
  • d58ac083f3 expose itemsize and nbytes as for numpy arrays (#284) Daniel Strobusch 2023-12-25 19:34:28 +01:00
  • a123c3c7d2 implement-batch-norm-layer (#217) __mo_san__ 2023-12-25 16:32:53 +01:00
  • 9e6b8c9f48 Refactor the reduction kernels (#277) Angelos Katharopoulos 2023-12-24 14:47:57 -08:00
  • 22fee5a383 Remove redundant assert in losses.py (#281) Zach Schillaci 2023-12-24 11:39:08 -05:00
  • 7365d142a3 random.uniform must respect dtype, even if lower precision than "low" (#280) Daniel Strobusch 2023-12-24 16:04:43 +01:00
  • 8b227fa9af fix no metal build (#276) Awni Hannun 2023-12-23 19:18:10 -08:00
  • 8c3da54c7d Fix failing test for log cosh loss (#275) Vidit Agarwal 2023-12-24 05:56:46 +05:30
  • acf1721b98 Corrected the example of value_and_grad (#274) Vidit Agarwal 2023-12-24 00:36:38 +05:30
  • f91f450141 Fix argmax returns documentation (#263) Finn Voorhees 2023-12-22 23:33:17 -05:00
  • cd3616a463 Revisit autorelease memory pools (#260) Ronan Collobert 2023-12-22 11:01:26 -08:00
  • d35fa1db41 Add Hinge, Huber and LogCosh losses (#199) Nicholas Santavas 2023-12-22 19:28:10 +01:00
  • e8deca84e0 Add dropout2d (#250) Justin Deschenaux 2023-12-22 17:02:29 +01:00
  • 8385f93cea Bumping the version (#256) v0.0.6 Angelos Katharopoulos 2023-12-21 18:33:14 -08:00
  • 2118c3dbfa fix (#255) Awni Hannun 2023-12-21 18:18:41 -08:00
  • a002797d52 A temporary fix (#254) Awni Hannun 2023-12-21 17:59:15 -08:00
  • 1d053e0d1d Fix the alibi test that was left unchanged (#252) Angelos Katharopoulos 2023-12-21 14:59:25 -08:00
  • 0aa65c7a6b Added ALiBi implementation (#232) Hazem Essam 2023-12-22 00:36:38 +02:00
  • 794feb83df support arange for bfloat16 (#245) Daniel Strobusch 2023-12-21 23:33:43 +01:00
  • 2c7df6795e Make sure that arrays are freed when saving (#247) Angelos Katharopoulos 2023-12-21 14:08:24 -08:00
  • b3916cbf2b Improve names of quantization arguments (#235) Angelos Katharopoulos 2023-12-20 16:53:53 -08:00
  • 57fe918cf8 Adds C++ and nn quantization utilities (#230) Angelos Katharopoulos 2023-12-20 14:17:38 -08:00
  • 4912ff3ec2 Add Lion optimizer (#209) Justin Deschenaux 2023-12-20 22:54:58 +01:00
  • f40d17047d Indexing bug (#233) Awni Hannun 2023-12-20 10:44:01 -08:00
  • 2807c6aff0 Implements divide for integer types and adds floor_divide op (#228) Angelos Katharopoulos 2023-12-19 20:12:19 -08:00
  • de892cb66c fix for non-macos build issue on cblas.h (#227) davidkoski 2023-12-19 17:01:59 -08:00
  • 37024d899c fixes for building with swiftpm (#225) davidkoski 2023-12-19 16:22:10 -08:00
  • 137f55bf28 fail early if readinto does not exist (#221) Diogo 2023-12-19 16:27:17 -05:00
  • e549f84532 Triplet Loss (#211) Emircan Erol 2023-12-19 23:37:12 +03:00
  • dfa9f4bc58 An initial quantized matmul implementation (#205) Angelos Katharopoulos 2023-12-18 23:18:57 -08:00
  • e6872a4149 Added linspace (#181) Abe Leininger 2023-12-18 19:57:55 -08:00
  • f4f6e17d45 Fix cross-attention (#210) Juarez Bochi 2023-12-18 15:27:27 -05:00
  • 4d4af12c6f Adds round op and primitive (#203) Angelos Katharopoulos 2023-12-18 11:32:48 -08:00
  • 477397bc98 Citation + Contributor acknowledgment section (#207) Awni Hannun 2023-12-18 10:07:00 -08:00
  • 18cca64c81 Add smoothed L1 loss and enhancements to cross entropy loss (#166) jojopuppet 2023-12-18 23:26:21 +08:00
  • 0e5807bbcb include optional (#202) Awni Hannun 2023-12-17 22:01:35 -08:00
  • 8eb56beb3a Added clip function (#159) Cyril Zakka, MD 2023-12-17 22:00:29 -06:00
  • ee0c2835c5 Docs updates (#198) Awni Hannun 2023-12-17 13:20:55 -08:00
  • 90d04072b7 fix build w/ flatten (#195) Awni Hannun 2023-12-17 11:58:45 -08:00
  • 52e1589a52 implemented Flatten Module (#149) __mo_san__ 2023-12-17 06:54:37 +01:00
  • eebd7c275d Add optimizers (AdaMax, AdaDelta, RMSprop) and ordering optimizer classes (#142) YUN, Junwoo 2023-12-17 13:43:15 +08:00
  • a67bbfe745 Update docs (#177) (#190) Austin Liu 2023-12-16 22:52:18 +08:00
  • 104c34f906 setite negative indexing bug (#189) Awni Hannun 2023-12-16 06:44:47 -08:00
  • dc2edc762c added tri / tril / triu (#170) Diogo 2023-12-15 20:30:34 -05:00
  • 2e02acdc83 add base kwarg to rope (#186) Awni Hannun 2023-12-15 16:47:59 -08:00
  • 83f266c44c Lazy metal_device_ initialization (#185) Ronan Collobert 2023-12-15 16:06:46 -08:00
  • f24200db2c accross -> across (#183) Víctor Aguilar 2023-12-15 13:46:50 -08:00
  • e28b57e371 Added mx.stack c++ frontend impl (#123) Jason 2023-12-14 16:21:19 -05:00
  • e5851e52b1 Add move and swap axis, and vmap for slice, concat, and gather (#158) Awni Hannun 2023-12-14 12:59:12 -08:00
  • f55908bc48 Added stubs for python files generated from C++ (#136) Diogo 2023-12-14 15:58:45 -05:00
  • b93c4cf378 Floor and Ceil (#150) Luca Arnaboldi 2023-12-14 19:00:23 +01:00