Skip to content

Pull requests: ml-explore/mlx

New pull request New
54 Open 2,842 Closed
54 Open 2,842 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

Clear CUDA errors once reported and raise eval errors in buffer protocol low priority
#4523 opened Sep 16, 2026 by aleroot Contributor Loading…
Add fused Metal kernels for fast.cross_entropy
#4520 opened Sep 16, 2026 by zsun6 Loading…
Fix ordering of complex64 with NaNs low priority
#4519 opened Sep 15, 2026 by louen Contributor Loading…
1
Improve pre-M5 metal memory usage for SDPA D512
#4518 opened Sep 15, 2026 by dhiltgen Contributor • Draft
fix(cpu): use size_t for contiguous binary op element counts (#4479) low priority
#4517 opened Sep 15, 2026 by Prudctual Loading…
1
1
Use the fast qmv kernel for affine outputs not divisible by 8 low priority
#4516 opened Sep 15, 2026 by wyanzhao Contributor Loading…
1
fix floor_divide low priority
#4515 opened Sep 15, 2026 by aaishwarymishra Contributor Loading…
1
Fix CUDA termination crash via deterministic worker teardown and lazy stream evaluation
#4514 opened Sep 15, 2026 by aleroot Contributor Loading…
1
Clamp slice bounds when a slice is combined with an array index low priority
#4513 opened Sep 15, 2026 by IshaanPotle Loading…
1
Fix floating-point constant precision in compiled kernels low priority
#4511 opened Sep 15, 2026 by Ryan11c Loading…
1
Support shapeless compilation of scan operations low priority
#4510 opened Sep 15, 2026 by keeeeenw Loading…
1
[CUDA] Add fp_gather_qmv to optimize gather_qmm await response This pull request is waiting for response from the author.
#4508 opened Sep 14, 2026 by dhiltgen Contributor Loading…
9
[CUDA] Add global scale support to gather_qmm await response This pull request is waiting for response from the author.
#4507 opened Sep 14, 2026 by dhiltgen Contributor Loading…
3
Improve metal memory usage for SDPA D256
#4505 opened Sep 14, 2026 by dhiltgen Contributor Loading…
Add fixed size unique operation
#4501 opened Sep 14, 2026 by louen Contributor Loading…
restoring perf regression for upsample linear mode align_corners=False
#4500 opened Sep 14, 2026 by sp4s-s Loading…
Add Metal SDPA support for D96/V64
#4499 opened Sep 14, 2026 by wyanzhao Contributor Loading…
Add missing defaults to tri, tril, triu, gather_mm signatures low priority
#4492 opened Sep 12, 2026 by rishabhsai Loading…
added pos support low priority
#4488 opened Sep 10, 2026 by aaishwarymishra Contributor Loading…
1
Improve metal memory usage for SDPA D512
#4487 opened Sep 10, 2026 by dhiltgen Contributor Loading…
Raise IndexError for out of bounds axes low priority
#4484 opened Sep 9, 2026 by devangpratap Loading…
1
3
Use NAX attention for short causal D256 prefill await response This pull request is waiting for response from the author.
#4476 opened Sep 8, 2026 by wyanzhao Contributor • Draft
3
Refactor: Extract set_output Function from utils.h await response This pull request is waiting for response from the author. low priority
#4467 opened Sep 5, 2026 by JasonHonKL Contributor Loading…
4
Support uneven tensor parallel sharding in distributed linear layers low priority
#4465 opened Sep 4, 2026 by twallgren Loading…
fix: keep single-element strided slice region round-tripping await verification This pull request is non-trivial and requires a human expert to verify its correctness. low priority
#4446 opened Sep 1, 2026 by axiom-of-choice Contributor Loading…
1
ProTip! Add no:assignee to see everything that’s not assigned.