Search results

Error when using mlx-community/gemma-4-e4b-it-4bit

Status: Open.

#1242 In ml-explore/mlx-lm;

· mikeazo opened on May 3, 2026

[Enhancement] Native Node.js bindings for mlx-lm — eliminate Python subprocess dependency

Status: Open.

#1237 In ml-explore/mlx-lm;

· jahbini opened on May 2, 2026

Bug report: "tool_parsers/qwen3_coder.py:79 crashes on ISO 8601 dates

Status: Open.

#1236 In ml-explore/mlx-lm;

· zubair-akbar opened on May 2, 2026

Feature request: idle-unload-timeout

Status: Open.

#1235 In ml-explore/mlx-lm;

· zubair-akbar opened on May 2, 2026

Add model support for DeepSeek-V4 (deepseek_v4) and Ling-2.6-flash (bailing_hybrid)

Status: Open.

#1233 In ml-explore/mlx-lm;

· Keenni opened on Apr 30, 2026

Feature Request: Add native support to skip or block `` reasoning content

Status: Open.

#1222 In ml-explore/mlx-lm;

· NeoInBJ opened on Apr 28, 2026

enable_thinking injection breaks deepseek_v32 chat_template

Status: Open.

#1221 In ml-explore/mlx-lm;

· blightbow opened on Apr 28, 2026

Mixed-bit Qwen3_5 multimodal quants fail to load — per-tensor quantization overrides not remapped through Model.sanitize

Status: Open.

#1214 In ml-explore/mlx-lm;

· sammcj opened on Apr 27, 2026

Gemma 4 attention-shape divergence between mlx-lm and mlx-swift-lm breaks fused-checkpoint round-trip

Status: Open.

#1210 In ml-explore/mlx-lm;

· kr1s0404 opened on Apr 27, 2026

v0.31.3: load_default → snapshot_download → thread_map crashes mlx_lm.server in a tight loop ("cannot schedule new futures after interpreter shutdown")

Status: Open.

#1208 In ml-explore/mlx-lm;

· twinsgeeks opened on Apr 27, 2026

LoRA training crashes on first backward pass with Qwen3.5-9B on M5 Max (applegpu_g17s)

Status: Open.

#1206 In ml-explore/mlx-lm;

· sivavenkatay opened on Apr 26, 2026

Skip lm_head on non-final pipeline-parallel ranks?

Status: Open.

#1203 In ml-explore/mlx-lm;

· adurham opened on Apr 26, 2026