Commit Graph

  • 2fdf9eb535 Fix ternary for large arrays (#1359) Awni Hannun 2024-08-26 11:22:27 -07:00
  • 860d3a50d7 fix extension metal library finding (#1361) Awni Hannun 2024-08-26 09:18:50 -07:00
  • d1183821a7 int() and float() for mx.array (#1360) Alex Barron 2024-08-25 20:41:44 -07:00
  • 8081df79be Fix boolean all reduce bug (#1355) v0.17.1 Angelos Katharopoulos 2024-08-24 10:09:32 -07:00
  • 64bec4fad7 Chore: update pre-commit hooks (#1353) Nripesh Niketan 2024-08-24 19:16:36 +05:30
  • b96e105244 Add grid_sample example to metal_kernel docs (#1352) Alex Barron 2024-08-23 18:24:16 -07:00
  • 3b4d5484c7 Bump extension MLX version (#1350) Awni Hannun 2024-08-23 12:38:34 -07:00
  • 684e11c664 patch (#1347) v0.17.0 Alex Barron 2024-08-23 10:42:02 -07:00
  • b57a52813b Further reduction tuning (#1349) Angelos Katharopoulos 2024-08-23 10:35:25 -07:00
  • da8deb2b62 fix bug with multiple attributes (#1348) Alex Barron 2024-08-23 10:06:15 -07:00
  • 98b6ce3460 Refactor reductions and fix scatter atomics for large sizes (#1300) Awni Hannun 2024-08-22 16:03:31 -07:00
  • f9e00efe31 fix nanobind and stub gen in circle (#1346) Awni Hannun 2024-08-22 14:07:27 -07:00
  • 0fd2a1f4b0 Custom Metal Kernels from Python (#1325) Alex Barron 2024-08-22 13:46:29 -07:00
  • df3233454d 2d gather specialization (#1339) Awni Hannun 2024-08-22 10:48:24 -07:00
  • 82db84b899 bump nanobind + fix extension (#1344) Awni Hannun 2024-08-21 16:05:07 -07:00
  • 8ae751d3da fix io (#1343) Awni Hannun 2024-08-21 13:14:46 -07:00
  • d40e76809f Fix rope (#1340) Awni Hannun 2024-08-20 17:37:52 -07:00
  • bb1b76d9dc RoPE with frequencies as optional input (#1337) Awni Hannun 2024-08-19 18:30:50 -07:00
  • 9d26441224 Fix contiguity check (#1336) Angelos Katharopoulos 2024-08-19 16:05:06 -07:00
  • f12f24a77c fix compiling with space in paths (#1332) Awni Hannun 2024-08-15 16:39:24 -07:00
  • ae5b5cabfd Fix optimizer reloading from checkpoint (#1329) Awni Hannun 2024-08-15 07:33:23 -07:00
  • d0630ffe8c Read arrays from files faster (#1330) Awni Hannun 2024-08-14 20:09:56 -07:00
  • 99bb7d3a58 GPU mx.sign for complex64 (#1326) Alex Barron 2024-08-14 07:54:53 -07:00
  • 63ae767232 fix transformer (#1327) Awni Hannun 2024-08-13 16:04:26 -07:00
  • eaaea02010 Add isfinite (#1318) Awni Hannun 2024-08-13 14:49:28 -07:00
  • a098bc92e0 Fix: Preserve input dtype in Dropout layer output (#1323) Bhargav Yagnik 2024-08-13 14:54:21 -04:00
  • 1086dc4db0 patch (#1320) v0.16.3 Awni Hannun 2024-08-12 16:13:33 -07:00
  • 19fb69e2ed Add memory_efficient_threshold kwarg to sdpa kernel (#1319) Brian Keene 2024-08-12 15:57:09 -04:00
  • 9231617eb3 Move to nanobind v2 (#1316) v0.16.2 Awni Hannun 2024-08-08 17:17:46 -07:00
  • 32668a7317 CPU mx.linalg.cholesky_inverse and mx.linalg.tri_inv (#1307) Alex Barron 2024-08-08 15:18:02 -07:00
  • 780c197f95 Fix test tolerance and patch bump (#1315) Angelos Katharopoulos 2024-08-08 14:51:09 -07:00
  • eb8819e91e Revert variance to be numerically stable (#1314) Angelos Katharopoulos 2024-08-08 13:35:02 -07:00
  • 30bbea2f08 Add gemv masked to JIT plus some fixes (#1310) Awni Hannun 2024-08-07 13:38:07 -07:00
  • 635ccd9e25 Add "edge" mode to mx.pad (#1309) Alex Barron 2024-08-06 11:23:10 -07:00
  • 8c9f0278b9 Add vmap to scatter (#1200) nicolov 2024-08-06 05:12:27 +02:00
  • 58d0e199e1 add bfloat conv for windograd (#1306) Awni Hannun 2024-08-05 15:51:13 -07:00
  • 10b5835501 fix creating array from bf16 tensors in jax / torch (#1305) Awni Hannun 2024-08-01 16:20:51 -07:00
  • 6c8dd307eb faster group norm (#1304) Awni Hannun 2024-08-01 12:49:23 -07:00
  • 43ffdab172 fix rope and random (#1301) Awni Hannun 2024-07-31 16:18:25 -07:00
  • 40b6d67333 Fixes for large arrays with a few ops (#1299) Awni Hannun 2024-07-30 17:18:39 -07:00
  • c52d1600f0 Fused Affine Quantize/Dequantize ops (#1282) Alex Barron 2024-07-29 15:11:38 -07:00
  • aa1d6cadad Fix docs latex build and nits (#1297) Awni Hannun 2024-07-29 11:44:06 -07:00
  • 6e06e3a904 feat: Added "tanh" option to GELU approximation (#1268) Atakan Tekparmak 2024-07-28 09:07:56 +02:00
  • 8cfb9fc0b8 Update requirements.txt (#1291) Yaroslav 2024-07-26 21:59:52 +02:00
  • 7b456fd2c0 Array api (#1289) Awni Hannun 2024-07-26 10:40:49 -07:00
  • e9e53856d2 patch bump (#1287) v0.16.1 Awni Hannun 2024-07-25 11:42:09 -07:00
  • 5029894662 [Issue #1187] Add nan_to_num function initial attempt (#1247) Anton Belov 2024-07-25 17:57:37 +01:00
  • baf9fa5f42 Einsum (#1269) Awni Hannun 2024-07-25 09:36:44 -07:00
  • 7f914365fd Fix GPU sort for large arrays (#1285) Jagrit Digani 2024-07-24 14:37:10 -07:00
  • ebd7135b50 Improve stability of BCE loss calculation for input probabilities close to or exactly 0 or 1 (#1280) Paul Paczuski 2024-07-24 11:38:22 -04:00
  • 50eff6a10a Implement sampling from laplace distribution. (#1279) fgranqvist 2024-07-24 15:15:37 +02:00
  • c34a5ae7f7 Fix bfloat16 Hadamard (#1283) Alex Barron 2024-07-23 14:54:43 -07:00
  • e2aa6ec8ae some fixes (#1281) Awni Hannun 2024-07-23 11:49:05 -07:00
  • 6768c6a54a Adding missing type hints (#1243) toji 2024-07-23 19:59:38 +05:30
  • 6307d166eb Fix overflow / underflow handling for expm1f (#1278) Tim Gymnich 2024-07-23 16:29:06 +02:00
  • 1fba87b0df Fix leak with multi-output primitives (#1274) Awni Hannun 2024-07-23 06:34:18 -07:00
  • df124e018a fix gguf (#1273) Awni Hannun 2024-07-18 07:35:35 -07:00
  • 2f83d6e4b7 Do not release buffers on exit (#1142) Cheng 2024-07-16 07:12:24 +09:00
  • 987785d8d7 Fix typo and missing header (#1266) Feng Shijie 2024-07-15 23:20:24 +08:00
  • 8c01a7893b minor fix in optimizer + docs (#1264) Awni Hannun 2024-07-12 12:18:02 -07:00
  • 218047c75a docs fixes (#1263) Awni Hannun 2024-07-11 15:59:07 -07:00
  • d0da74209b version bump (#1260) v0.16.0 Alex Barron 2024-07-11 11:17:55 -07:00
  • 5c1fa64fb0 Custom transforms (#1246) Angelos Katharopoulos 2024-07-10 18:00:01 -07:00
  • a3c287354f Fast Hadamard Transform (#1249) Alex Barron 2024-07-09 20:39:01 -07:00
  • 03cf033f82 Fix reshape copy bug (#1253) Angelos Katharopoulos 2024-07-07 21:37:00 -07:00
  • bdb36c9a63 add zero vjps for bitwise ops and gather w.r.t. index (#1256) Alex Barron 2024-07-07 21:34:59 -07:00
  • 20bb301195 CPU binary reduction + Nits (#1242) Awni Hannun 2024-06-28 13:50:42 -07:00
  • d6383a1c6a version bump (#1239) v0.15.2 Awni Hannun 2024-06-27 10:43:13 -07:00
  • b05bcfd27f Fixes segfault when compiling checkpointed functions (#1235) Angelos Katharopoulos 2024-06-26 16:14:45 -07:00
  • 2615660e62 Fix strided sort bug (#1236) Alex Barron 2024-06-26 14:32:11 -07:00
  • 5b0af4cdb1 fix donation condition for compilation (#1237) Awni Hannun 2024-06-26 09:04:05 -07:00
  • 8c2e15e6c8 Accelerate import updates for iOS (#1227) Jagrit Digani 2024-06-26 09:01:50 -07:00
  • 56c8a33439 Get metal version from xcode (#1228) Awni Hannun 2024-06-26 07:02:11 -07:00
  • 4eef1e8a3e fix typo (#1215) David Koski 2024-06-24 13:36:35 -07:00
  • 95d11bda06 Fix NumPy 2.0 pickle test (#1221) Alex Barron 2024-06-23 05:47:22 -07:00
  • af9079cc1f version bump (#1212) v0.15.1 Awni Hannun 2024-06-14 11:28:51 -07:00
  • 2d6cd47713 Masked gemv (#1211) Jagrit Digani 2024-06-14 09:52:26 -07:00
  • fe3167d7ea smaller CPU binary (#1203) Awni Hannun 2024-06-14 09:46:55 -07:00
  • 31e134be35 Build for macOS 15 (#1208) Awni Hannun 2024-06-13 13:31:44 -07:00
  • e84ba8056d only allow openmpi (#1209) Awni Hannun 2024-06-13 12:14:44 -07:00
  • f20e97b092 minor fixes (#1194) Fangjun Kuang 2024-06-13 13:06:49 +08:00
  • 934683088e Refactor JIT for unary/binary/ternary ops (#1206) Alex Barron 2024-06-12 14:22:12 -07:00
  • de2b9e7d0a Fix kernel deps to reduce build times (#1205) Awni Hannun 2024-06-12 11:17:39 -07:00
  • dd7d8e5e29 Add Quantized Ops to the JIT (#1204) Alex Barron 2024-06-12 09:47:12 -07:00
  • df964132fb fix scatter + test (#1202) Awni Hannun 2024-06-11 14:35:12 -07:00
  • 709ccc6800 install mpi for release build (#1199) Awni Hannun 2024-06-10 10:09:32 -07:00
  • cf236fc390 version (#1191) v0.15.0 Awni Hannun 2024-06-06 17:16:40 -07:00
  • 27d70c7d9d Feature complete Metal FFT (#1102) Alex Barron 2024-06-06 12:57:25 -07:00
  • 0e585b4409 Add docstring for scatter (#1189) nicolov 2024-06-06 20:51:25 +02:00
  • 0163a8e57a Add docs for the distributed namespace (#1184) Angelos Katharopoulos 2024-06-06 11:37:00 -07:00
  • 578842954c fix jit scan when output doesn't have primitive (#1190) Awni Hannun 2024-06-06 07:24:58 -07:00
  • 496315fe1d Fix scan (#1188) Awni Hannun 2024-06-05 14:21:58 -07:00
  • 0fe6895893 Fix the hard-shrink test (#1185) Angelos Katharopoulos 2024-06-04 16:22:56 -07:00
  • 0b7d71fd2f Add softmin, hardshrink, hardtanh (#1180) Nikhil Mehta 2024-06-04 15:48:18 -07:00
  • 83b11bc58d Fix Metal API validation for empty concat (#1183) Awni Hannun 2024-06-04 13:17:08 -07:00
  • 375a8bbdcc Add some internal GPU apis (#1177) Alex Barron 2024-06-04 09:24:26 -07:00
  • ea9090bbc4 Add view op (#1179) Awni Hannun 2024-06-04 08:05:27 -07:00
  • 81def6ac76 Fix benchmark (#1175) nicolov 2024-06-04 16:50:46 +02:00
  • 3de8ce3f3c In place all-reduce and forgiving init (#1178) Angelos Katharopoulos 2024-06-03 16:47:47 -07:00
  • 4d485fca24 Add defines include (#1176) Alex Barron 2024-06-03 09:50:10 -07:00