Skip to content

Pull requests: ml-explore/mlx

New pull request New
86 Open 2,373 Closed
86 Open 2,373 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Fix signed-integer overflow in convolution shape arithmetic
#3938 opened Jul 28, 2026 by eyupcanakman Contributor Loading…
Metal allocator: round buffer cache lookups to a size class so growing allocations get reused
#3937 opened Jul 28, 2026 by PhilipJohnBasile Loading…
Fix mx.random.randint precision loss beyond 2^24 (float32 domain sampling)
#3936 opened Jul 28, 2026 by PhilipJohnBasile Loading…
metal: reduce NVFP4 scales per 16-lane group
#3934 opened Jul 27, 2026 by jessegross Contributor Loading…
Fix crashes in the ring and jaccl distributed backends
#3933 opened Jul 27, 2026 by jzdziarski Loading…
3 of 4 tasks
Fix Metal column reduction grid overflow
#3931 opened Jul 27, 2026 by jasp-nerd • Draft
4 tasks done
1
Add CUDA 13 runtime headers
#3930 opened Jul 27, 2026 by jasp-nerd • Draft
4 tasks done
1
Fix CUDA batched GEMV grid overflow
#3929 opened Jul 27, 2026 by jasp-nerd • Draft
4 tasks done
1
Restore type stubs in frontend wheels
#3928 opened Jul 27, 2026 by jasp-nerd • Draft
4 tasks done
1
Add special functions (erf, erfc, i0, gammaln, digamma) as mlx.special
#3927 opened Jul 27, 2026 by priyanka25aug Loading…
4 tasks done
Add a tile-shape static_assert to tile_matmad_nax
#3924 opened Jul 26, 2026 by dwijenpatel Loading…
Add BitLinear: BitNet b1.58 quantization-aware training layer
#3923 opened Jul 26, 2026 by peterlodri-sec Loading…
4 of 5 tasks
Fix sorted gather_qmm NAX boundary handling
#3922 opened Jul 26, 2026 by PhilipJohnBasile Loading…
2
1
Reduce per-eval CPU overhead: flat degree map in eval_impl + gather identity-index cache
#3920 opened Jul 25, 2026 by spokvulcan Loading…
[Metal] Fused causal-mask + softmax for the SDPA ops fallback (bit-identical)
#3919 opened Jul 25, 2026 by spokvulcan Loading…
[Metal] Rows-per-expert-aware gather_qmm_rhs tile geometry
#3918 opened Jul 25, 2026 by spokvulcan Loading…
Clamp the runtime start in slice and slice_update so the window stays in bounds
#3917 opened Jul 25, 2026 by kapellirohith Loading…
1
Raise a clear error for an invalid quantization mode in nn layers
#3914 opened Jul 25, 2026 by ayaangazali Contributor Loading…
3 of 4 tasks
[Metal] Faster logsumexp for short rows via a simdgroup-per-row kernel
#3913 opened Jul 24, 2026 by qx555 Loading…
Fix fp quantized matmul corruption when the quantized dim is not a multiple of 32
#3912 opened Jul 24, 2026 by kapellirohith Loading…
3
Accumulate low-precision float reductions in float32 on the CPU
#3909 opened Jul 24, 2026 by v-code01 Loading…
2
Fix Upsample align_corners=False for fractional scale factors
#3908 opened Jul 24, 2026 by v-code01 Loading…
Accumulate low-precision float scans in float32 on the CPU
#3907 opened Jul 24, 2026 by v-code01 Loading…
Fix filter_and_map docstring argument order for filter_fn and is_leaf_fn
#3906 opened Jul 24, 2026 by ayaangazali Contributor Loading…
3 of 4 tasks
add runnable data parallel example
#3905 opened Jul 24, 2026 by tylergibbs1 Loading…
2 of 4 tasks
ProTip! What’s not been updated in a month: updated:<2026-06-28.