All issues

Search results

ReduceScatter autograd (jvp/vjp/vmap)

Status: Open.

#4405 In ml-explore/mlx;

· Goekdeniz-Guelmez opened on Aug 25, 2026

[BUG] NAX MoE simdgroup skipping in 0.32.2 triggers lost JACCL completion under heterogeneous TP

bug

distributed

Status: Open.

#4403 In ml-explore/mlx;

· jonathan308 opened on Aug 25, 2026

[BUG] CUDA fence host blocking with GPU-to-GPU cross-stream dependencies

Status: Open.

#4400 In ml-explore/mlx;

· strayberry opened on Aug 25, 2026

[BUG] Mixed slice + fancy indexing produces wrong shape/garbage data for out-of-range slice bounds

Status: Open.

#4399 In ml-explore/mlx;

· Adityaj0 opened on Aug 25, 2026

[BUG] Crash on ellipsis indexing with too many trailing indices

Status: Open.

#4398 In ml-explore/mlx;

· Adityaj0 opened on Aug 25, 2026

[BUG] Metal Kernel Compilation Issue on M2 macOS 15 (Sequoia)

Status: Open.

#4389 In ml-explore/mlx;

· marianovosolov opened on Aug 24, 2026

linalg.inv on CPU stream aborts the process on a singular matrix instead of raising

Status: Open.

#4386 In ml-explore/mlx;

· neuromechanist opened on Aug 23, 2026

[BUG] mx.var / mx.std wrong for offset data

Status: Open.

#4379 In ml-explore/mlx;

· JasonHonKL opened on Aug 23, 2026

[BUG] __setitem__ with a single Ellipsis fails

bug

low priority

python

Status: Open.

#4358 In ml-explore/mlx;

· ev-br opened on Aug 19, 2026

Support fusing element-wise operations into reductions

enhancement

Status: Open.

#4355 In ml-explore/mlx;

· sakupan102 opened on Aug 19, 2026

[BUG] mx.abs / mx.sqrt / mx.log on complex64 are silently wrong on GPU across a third of float32's range

bug

Status: Open.

#4344 In ml-explore/mlx;

· JasonHonKL opened on Aug 18, 2026

[CUDA] gather_qmm ~24x slower than equal-FLOPs quantized_matmul on GB10 (sm_121) — makes quantized MoE prefill unusable

cuda

performance

Status: Open.

#4339 In ml-explore/mlx;

· NORDIT opened on Aug 18, 2026