Commit Graph

  • 4e1e9520e1 Flatten and unflatten (#1692) Awni Hannun 2024-12-11 21:51:37 -08:00
  • 0bf19037ca Remove "using namespace mlx::core" in python/src (#1689) Cheng 2024-12-12 08:45:39 +09:00
  • f3dfa36a3a Fix x86 tests (#1691) Awni Hannun 2024-12-11 07:47:18 -08:00
  • 4f9b60dd53 Remove "using namespace mlx::core" in benchmarks/examples (#1685) Cheng 2024-12-12 00:08:29 +09:00
  • f76a49e555 ExpandDims primitive (#1687) Awni Hannun 2024-12-10 16:39:07 -08:00
  • 310ad8d9db Build OpenBLAS from source code for MSVC (#1674) Cheng 2024-12-11 09:14:44 +09:00
  • 56db268f47 Provide a pread implementation for MSVC (#1666) Cheng 2024-12-11 08:55:53 +09:00
  • 92ab6bdeb8 Fix shared library not exporting symbols on Windows (#1684) Cheng 2024-12-11 06:59:14 +09:00
  • 0070e360a1 Disable MSVC warnings (#1680) Cheng 2024-12-10 12:41:14 +09:00
  • 9df8fed046 Metal-cpp version bump (#1668) Amethyst Shen 2024-12-09 22:40:35 -05:00
  • a59fae040f Fix library output directory for MSVC (#1681) Cheng 2024-12-10 12:07:50 +09:00
  • 29a620cab2 No reshapes in quantized embedding (#1682) Awni Hannun 2024-12-09 18:57:38 -08:00
  • 87d7a2520e Use Py_ssize_t in python bindings (#1678) Cheng 2024-12-10 05:59:19 +09:00
  • 40c62c1321 Use int64 stride everywhere (#1671) Awni Hannun 2024-12-09 11:09:02 -08:00
  • 35b412c099 Fix compile hasher for string constants. (#1677) Awni Hannun 2024-12-09 09:26:18 -08:00
  • d0f471cff7 Using math defines requires switch in MSVC (#1665) Cheng 2024-12-09 01:16:28 +09:00
  • 6f316b8bf5 Use int64_t instead of ssize_t (#1673) Cheng 2024-12-08 13:10:44 +09:00
  • 7c10c93a1f Convert filesystem path to std::string explicitly (#1672) Cheng 2024-12-08 13:10:06 +09:00
  • d92ea094f1 Use && instead of and (#1663) Cheng 2024-12-08 11:26:39 +09:00
  • 6ae5423b4a Do not pass integers to isnan (#1664) Cheng 2024-12-08 11:26:23 +09:00
  • 9635cffdc8 Include io.h in MSVC for IO functions (#1661) Cheng 2024-12-08 11:26:06 +09:00
  • 96986fb362 Use auto* for pointers (#1662) Cheng 2024-12-08 11:25:40 +09:00
  • 3ceb341a75 Use correct complex type for MSVC (#1660) Cheng 2024-12-08 11:25:22 +09:00
  • 50fa705125 patch bump (#1656) v0.21.1 Awni Hannun 2024-12-06 13:16:19 -08:00
  • 69a2991614 allow compiling lambdas in C++ (#1650) Awni Hannun 2024-12-06 13:13:21 -08:00
  • fd3377dd1f Support bias correction in Adam and AdamW optimizers (#1640) mt_caret 2024-12-07 05:13:34 +09:00
  • d0b6cb0425 More primitives for compiling with shapeless (#1653) Awni Hannun 2024-12-06 11:29:18 -08:00
  • 95c4a2e3af add back conditionaltype (#1655) Alex Barron 2024-12-06 11:12:01 -08:00
  • bc2a29f033 fix (#1654) Awni Hannun 2024-12-06 10:48:58 -08:00
  • 82a956c1d9 fix test q-sdpa Alex Barron 2024-12-06 10:26:54 -08:00
  • 3bb5b4a302 Chore: Add default language in pre-commit and bump hooks (#1652) Nripesh Niketan 2024-12-06 15:54:29 +00:00
  • 769704653a cpu fallback Alex Barron 2024-12-06 01:22:50 -08:00
  • c89ddf62b4 add checks Alex Barron 2024-12-06 01:09:00 -08:00
  • 3507c104a5 add test Alex Barron 2024-12-06 00:45:01 -08:00
  • 12a4d89a7c working qsdpa Alex Barron 2024-12-06 00:14:24 -08:00
  • fc88fd9097 Shape and Strides 1 / N (#1645) Awni Hannun 2024-12-05 12:53:43 -08:00
  • c5b0928c1f fix fallback (#1646) Awni Hannun 2024-12-05 11:59:53 -08:00
  • e047fd977d compile changes if stream changes (#1644) Awni Hannun 2024-12-03 14:37:44 -08:00
  • 9d40e521d7 Stop matrix copies with new attention kernel (#1639) Jagrit Digani 2024-12-02 14:12:38 -08:00
  • 1445dcaa60 let class predicate specify quantization parameters (#1638) Alex Barron 2024-12-02 14:09:28 -08:00
  • e4eeb4e910 Added missing unordered_map includes (#1635) Jesper Stemann Andersen 2024-12-02 16:03:03 +01:00
  • aa86876813 fix transformer decoder post norm LN (#1637) Awni Hannun 2024-12-02 07:02:17 -08:00
  • 974bb54ab2 CMake: Enabled using Accelerate on x86_64 / x64 (#1625) Jesper Stemann Andersen 2024-11-28 19:55:45 +01:00
  • 9bc2183a31 docs: update device.cpp (#1632) Ikko Eltociear Ashimine 2024-11-28 13:58:26 +09:00
  • d4b222b6d3 Fix some leaks and races (#1629) Awni Hannun 2024-11-27 20:01:20 -08:00
  • af2af818a6 Enables build for *-linux-musl (#1627) Jesper Stemann Andersen 2024-11-27 22:14:24 +01:00
  • 698e63a608 CMake: Build with dlfcn-win32 to have dlopen etc. on win32 (#1628) Jesper Stemann Andersen 2024-11-27 22:14:13 +01:00
  • 211411faf2 fix large ops (#1620) Awni Hannun 2024-11-24 09:17:10 -08:00
  • bb303c45a5 version (#1617) v0.21.0 Awni Hannun 2024-11-22 12:00:03 -08:00
  • 6f7986d592 Cleaner qmv/qvm (#1616) Alex Barron 2024-11-22 11:14:08 -08:00
  • 7cbb4aef17 Doc fix (#1615) Awni Hannun 2024-11-22 11:12:25 -08:00
  • 02bec0bb6d Matrix Attention kernel (#1610) Jagrit Digani 2024-11-22 10:34:05 -08:00
  • c79f6a4a8c 3 and 6 bit quantization (#1613) Alex Barron 2024-11-22 10:22:13 -08:00
  • 0c5eea226b Reduce specializations (#1607) Awni Hannun 2024-11-21 19:53:00 -08:00
  • dcca0d7477 contiguous op / prim (#1612) Awni Hannun 2024-11-21 19:51:49 -08:00
  • 0d5e7716ad fix typo: accross -> across (#1609) Cocoa 2024-11-20 23:30:51 +00:00
  • d8c824c594 Formatting fixes (#1606) Angelos Katharopoulos 2024-11-20 15:30:36 -08:00
  • cb431dfc9f Adds 3D pooling (#1526) Saanidhya 2024-11-19 19:45:24 -05:00
  • 61d787726a Fix view scalar bug segfault (#1603) Awni Hannun 2024-11-19 10:54:05 -08:00
  • 5e89aace9b Fix concatenate vmap (#1600) Angelos Katharopoulos 2024-11-19 10:44:04 -08:00
  • 2af7e8a9a6 fix cmake version (#1601) Awni Hannun 2024-11-19 08:45:05 -08:00
  • 2419edd5b2 Faster indexing math in a few kernels (#1589) Awni Hannun 2024-11-18 19:52:00 -08:00
  • bf481e8e5d Fix sibling leak (#1590) Awni Hannun 2024-11-18 19:17:01 -08:00
  • 9d7fa6b8e6 Use osx deployment target to pick Metal version (#1595) Awni Hannun 2024-11-18 19:16:49 -08:00
  • 073076ac7d 2-Pass Sdpa Inference Kernel (#1597) Angelos Katharopoulos 2024-11-18 17:31:53 -08:00
  • 9bd03dd9b4 More buffer donation with no-ops (#1591) Awni Hannun 2024-11-18 08:35:41 -08:00
  • 6931f84412 fix dispatch threads for a few kernels (#1594) Awni Hannun 2024-11-18 08:35:25 -08:00
  • 16ec0556a0 Allocate raw JSON metadata buffer on the heap, and limit its size (#1596) xnorai 2024-11-18 07:22:51 -08:00
  • 610af352d4 Dispatch bf16 at run time when using the JIT (#1584) Awni Hannun 2024-11-15 16:54:36 -08:00
  • b35f1e3c9c fix donation in sdpa (#1587) Awni Hannun 2024-11-13 17:21:13 -08:00
  • dfa0b9aab4 Cpu fast quantize (#1578) Awni Hannun 2024-11-08 20:10:39 -08:00
  • a4c47b0276 OOB QMV fix (#1579) Alex Barron 2024-11-08 17:59:45 -08:00
  • 111fefd5e9 Fix OOB access in qmv (#1577) Alex Barron 2024-11-08 15:41:30 -08:00
  • c1fe1ef081 Bfs width limit (#1568) Awni Hannun 2024-11-08 15:00:46 -08:00
  • 8c34c9dac4 throw for invalid case and remove test (#1575) Awni Hannun 2024-11-08 12:04:03 -08:00
  • 91c0277356 fix per-example mask + docs in sdpa (#1574) Awni Hannun 2024-11-08 11:51:15 -08:00
  • 9f0d5c12fc Fully wrap the command encoder (#1572) Awni Hannun 2024-11-08 11:50:21 -08:00
  • 59247c2b62 add groups in conv2d (#1569) Awni Hannun 2024-11-07 13:57:53 -08:00
  • 9a3842a2d9 fix (#1566) Awni Hannun 2024-11-06 17:10:33 -08:00
  • 726dbd9267 v0.20.0 (#1565) v0.20.0 Alex Barron 2024-11-05 12:37:57 -08:00
  • 54f05e7195 Fix gather vmap (#1563) Awni Hannun 2024-11-05 11:29:20 -08:00
  • 26be608470 Add split_k qvm for long context (#1564) Alex Barron 2024-11-05 11:25:19 -08:00
  • 248431eb3c Reductions update (#1351) Angelos Katharopoulos 2024-11-04 22:25:16 -08:00
  • 76f275b4df error in rms for wrong size (#1562) Awni Hannun 2024-11-04 13:24:02 -08:00
  • f1951d6cce Use fewer barriers (#1561) Awni Hannun 2024-11-04 10:26:49 -08:00
  • 62f297b51d Sdpa fix (#1558) Angelos Katharopoulos 2024-11-02 21:25:46 -07:00
  • 09bc32f62f No extra reshape (#1557) Awni Hannun 2024-11-02 19:07:20 -07:00
  • 46d8b16ab4 Fix vmap example in docs (#1556) Chris Offner 2024-11-03 01:44:14 +01:00
  • 42533931fa Fix typo "it's" -> "its" (#1555) Chris Offner 2024-11-02 14:06:34 +01:00
  • 9bd3a7102f add python 3.13 to circle (#1553) Awni Hannun 2024-11-01 20:55:35 -07:00
  • 9e516b71ea Add dispatchThreads to custom kernel doc (#1551) Alex Barron 2024-11-01 13:07:48 -07:00
  • eac961ddb1 patch (#1550) v0.19.3 Awni Hannun 2024-10-31 16:10:14 -07:00
  • 57c6aa7188 fix multi output leak (#1548) Awni Hannun 2024-10-31 09:32:01 -07:00
  • cde5b4ad80 patch (#1546) v0.19.2 Awni Hannun 2024-10-30 19:31:22 -07:00
  • 4f72c66911 improvements to scatter / gather (#1541) Awni Hannun 2024-10-30 19:30:54 -07:00
  • 960e3f0f05 Gemm update (#1518) Jagrit Digani 2024-10-30 19:30:28 -07:00
  • 884af42da2 Fix thread group for large arrays (#1543) Awni Hannun 2024-10-30 16:25:12 -07:00
  • 048fabdabd Fix vmap constant output size (#1524) Alex Barron 2024-10-30 16:16:53 -07:00
  • 917252a5a1 Add favicon to docs (#1545) Léo 2024-10-30 21:54:13 +01:00
  • 1a992e31e8 Skip using Residency sets in VMs (#1537) Carlo Cabrera 2024-10-30 10:37:23 +08:00