Cheng
|
d01b83dfe7
|
Use nb::ndarray for checking arrays (#3283)
|
2026-03-25 10:44:54 +09:00 |
|
Sheldon Aristide
|
1b1c56352a
|
Fix moved-from shape bug in broadcast_arrays causing vmap bus error (#3310)
|
2026-03-24 17:02:31 -07:00 |
|
Sheldon Aristide
|
81530c261b
|
Implement Pad::vmap (#3304)
Co-authored-by: Angelos Katharopoulos <a_katharopoulos@apple.com>
|
2026-03-24 15:02:31 -07:00 |
|
Cheng
|
70a0da6fca
|
Use thread local storage for frontend compile cache (#3280)
|
2026-03-20 07:44:45 +09:00 |
|
Long Yixing
|
82809ebd12
|
Fix sort NaN handling for float16 and bfloat16 (#3269)
|
2026-03-19 15:19:41 -07:00 |
|
AN Long
|
5fa1a8d59f
|
Support indexing with any type which implmented __index__ (#3210)
|
2026-03-19 15:19:08 -07:00 |
|
Cheng
|
21c11fc9b0
|
Create default random key lazily (#3278)
|
2026-03-19 20:22:52 +09:00 |
|
Angelos Katharopoulos
|
45af0df90b
|
Fix repr of conv layers (#3275)
|
2026-03-18 22:47:38 -07:00 |
|
Cheng
|
75f74ea9bc
|
Fix building with CUDA toolkit 13.2 (#3273)
|
2026-03-19 08:31:44 +09:00 |
|
Angelos Katharopoulos
|
7bc61cceed
|
Slice update with operation (#3266)
|
2026-03-18 06:18:02 -07:00 |
|
Ihar Hrachyshka
|
e353be8235
|
tests: harden memory leak check in test_siblings_without_eval (#3088)
Signed-off-by: Ihar Hrachyshka <ihar.hrachyshka@gmail.com>
|
2026-03-17 16:03:10 +09:00 |
|
mm65x
|
f226eeec9e
|
Fix nn.GRU skipping bhn bias when hidden is None (#3252)
Co-authored-by: mm65x <mm65x@users.noreply.github.com>
|
2026-03-16 13:28:14 -07:00 |
|
mm65x
|
505fc9850d
|
Fix comparison op JVP returning bool tangents instead of input dtype (#3253)
|
2026-03-16 10:57:28 -07:00 |
|
Long Yixing
|
0bdbfdb838
|
[CUDA] Implement MaskedScatter (#3151)
|
2026-03-15 10:33:55 +09:00 |
|
Lucas Newman
|
5d1700493a
|
[CUDA] Add FFT support (#3243)
|
2026-03-14 21:02:19 +09:00 |
|
Long Yixing
|
a9573f92f6
|
[CUDA] Implement SegmentedMM (#3238)
|
2026-03-11 13:31:43 -07:00 |
|
Anastasiia Filippova
|
e1e1399e1b
|
Hybrid sharding (#3194)
|
2026-03-10 11:47:25 +01:00 |
|
Cheng
|
9d03a1b0d9
|
[CUDA] Support 3/5/6-bit quants in QMV (#3236)
|
2026-03-10 19:09:48 +09:00 |
|
Michelle DiMarco
|
d2702a4fc1
|
Fix non-strict module update with extra weights (#3214)
Co-authored-by: Michelle DiMarco <m_dimarco@apple.com>
|
2026-03-09 22:03:50 -07:00 |
|
Dan Anderson
|
572e0a4ac3
|
Validate dims in rope (#3230)
Co-authored-by: KD2YCU <me@kd2ycu.com>
|
2026-03-09 21:48:45 -07:00 |
|
Dan Anderson
|
a25399cbd4
|
Validate num_splits in split (#3234)
Co-authored-by: KD2YCU <me@kd2ycu.com>
|
2026-03-09 21:47:08 -07:00 |
|
Dan Anderson
|
db487f3649
|
PR #3220 LayerNorm VJP returns zeros_like(weight) instead of zeros_like(bias placeholder) (#3231)
Co-authored-by: KD2YCU <me@kd2ycu.com>
|
2026-03-09 17:06:12 -07:00 |
|
Angelos Katharopoulos
|
d06c3c8936
|
Improve mlx.distributed_config (#3199)
|
2026-03-09 13:17:51 -07:00 |
|
Long Yixing
|
be872ebdef
|
[CUDA] implement Hadamard transform (#3179)
|
2026-03-05 09:34:19 +01:00 |
|
willem adnet
|
8cd377b7db
|
Add the bartlett function (#3155)
|
2026-03-03 11:40:54 -08:00 |
|
AN Long
|
1ce7118303
|
Fix ref leak in mx.save/load with file like object (#3187)
|
2026-03-02 16:55:29 -08:00 |
|
Anastasiia Filippova
|
72e04f7fb7
|
[CUDA] Fsdp (easy) (#3130)
|
2026-03-01 23:29:09 +01:00 |
|
Robert Johansson
|
c8536f5248
|
Fix compile_fuse broadcast split aliasing bug (#3166)
Co-authored-by: Angelos Katharopoulos <a_katharopoulos@apple.com>
|
2026-02-26 18:17:28 -08:00 |
|
Anastasiia Filippova
|
4e00919e5c
|
[CUDA][NCCL] group split (#3172)
|
2026-02-26 09:26:20 +01:00 |
|
willem adnet
|
a8ba5ac3e0
|
Implement mlx.core.blackman (#3136)
|
2026-02-25 13:42:40 -08:00 |
|
Gleb Sterkin
|
1d8d693d08
|
[Metal] Add implicit matmul pathway for mx.conv3d (#3147)
Co-authored-by: Gleb Sterkin <g_sterkin@apple.com>
Co-authored-by: Angelos Katharopoulos <a_katharopoulos@apple.com>
|
2026-02-23 17:52:50 -08:00 |
|
Alex Skryl
|
f2f2d16451
|
Export: preserve Dtype state values in export callback arguments (#3145)
Co-authored-by: Awni Hannun <awni@apple.com>
|
2026-02-19 08:07:28 -08:00 |
|
Anastasiia Filippova
|
06305022ab
|
Tensor scale nvfp4 (#3022)
|
2026-02-18 11:19:26 +01:00 |
|
willem adnet
|
360639c2df
|
Add the hamming window function (#3135)
|
2026-02-17 00:56:05 -08:00 |
|
willem adnet
|
3bbe87e6dc
|
Add hanning window function (#3124)
|
2026-02-16 09:44:49 -08:00 |
|
vskiwi
|
e226af720e
|
Propagate quantization mode in quantized layers (#3133)
|
2026-02-15 18:33:13 -08:00 |
|
Cheng
|
72e94c81e1
|
[CUDA] Attention sinks in cuDNN SDPA (#3118)
|
2026-02-11 16:46:39 +09:00 |
|
Cheng
|
54bb3eea42
|
[CUDA] Use cuDNN SDPA for decoding when using fixed-size KV cache (#3113)
|
2026-02-10 09:15:45 +09:00 |
|
Anastasiia Filippova
|
5e018de4e5
|
Quantize module to QQLinear (#3106)
|
2026-02-09 14:35:17 -08:00 |
|
Cheng
|
566bc16b7c
|
Cleanup test_fast_sdpa.py (#3112)
|
2026-02-08 19:04:24 +09:00 |
|
Manuel Candales
|
90e38f7b93
|
Fix qmv_impl for small N (#3096)
|
2026-02-05 17:33:36 -08:00 |
|
Angelos Katharopoulos
|
ceea571490
|
JACCL update (#3094)
|
2026-02-05 15:16:07 -08:00 |
|
Cheng
|
8ef539522c
|
Fix failing python tests on Windows (#3076)
|
2026-01-30 17:50:18 +09:00 |
|
Awni Hannun
|
fcbdd05022
|
More useful error for large indices (#3079)
|
2026-01-29 13:02:39 -08:00 |
|
atharva
|
590b4f1c16
|
Fix ALiBi slopes for non-power-of-2 num_heads (#3071)
|
2026-01-29 07:23:11 -08:00 |
|
Anri Lombard
|
0c6a895ed7
|
Use lower-right causal mask alignment consistently (#2967)
Co-authored-by: Awni Hannun <awni.hannun@gmail.com>
|
2026-01-28 17:15:14 -08:00 |
|
stef
|
c86a9bced1
|
[Docs] Simple example of using MLX distributed (#2973)
Co-authored-by: Awni Hannun <awni@apple.com>
|
2026-01-28 17:14:56 -08:00 |
|
Awni Hannun
|
4912cc47c2
|
Fp qmv (#2984)
|
2026-01-27 06:33:06 -08:00 |
|
Daniel Hiltgen
|
a828e769be
|
GPU discovery (#3055)
Co-authored-by: Angelos Katharopoulos <a_katharopoulos@apple.com>
|
2026-01-26 09:54:13 -08:00 |
|
Nripesh Niketan
|
b6aa03e5b8
|
Update pre-commit hooks and versions for clang-format, black, and isort (#3059)
|
2026-01-26 06:57:04 -08:00 |
|