Skip to content

Pull requests: ml-explore/mlx-lm

New pull request New
105 Open 621 Closed
105 Open 621 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

Auto-discover tool-call markers from tokenizer config fields
#1163 opened Apr 18, 2026 by michaelstingl Loading…
6 tasks done
feat(nemotron_h): add Multi-Token Prediction (MTP) module
#1161 opened Apr 16, 2026 by Thump604 Loading…
Add reasoning → tool state machine transition
#1160 opened Apr 16, 2026 by christiangenco Loading…
Fix Gemma 4 KV-shared layers creating unused projections
#1158 opened Apr 15, 2026 by glyphVault Loading…
5 tasks done
feature: dynamic quantized model support
#1155 opened Apr 15, 2026 by dsrenesanse Loading…
Feat/mamba mlx kernels
#1153 opened Apr 15, 2026 by Gal-bloch Loading…
9 of 11 tasks
1
feat: Add KL Divergence command
#1146 opened Apr 13, 2026 by spicyneuron Contributor Loading…
Add TurboQuantKVCache: data-oblivious 2-4 bit KV cache compression
#1144 opened Apr 12, 2026 by Smilefounder • Draft
3 tasks done
1
fix(gemma4): return [] instead of raising on empty tool-call match
#1142 opened Apr 10, 2026 by gofastercloud Loading…
1
[Fix] Add alias for served model
#1140 opened Apr 9, 2026 by austin362667 Loading…
3
Add pipeline parallel support for Qwen3 MoE and MiniMax models
#1138 opened Apr 9, 2026 by qubitcontracting Loading…
Pipeline parallel: memory-proportional splitting and inference sync
#1137 opened Apr 9, 2026 by qubitcontracting Loading…
Add RAG example using mlx-lm hidden state embeddings
#1130 opened Apr 8, 2026 by ManjushaMotamarry Loading…
feat(tuner): support loading PEFT/Unsloth LoRA adapters in load_adapters()
#1120 opened Apr 7, 2026 by YUGOROU Loading…
1
fix: honor --prompt-cache-bytes in sequential serve mode
#1118 opened Apr 7, 2026 by Jw983cam Loading…
2
fix: BatchRotatingKVCache.merge() shape mismatch with different fill levels
#1116 opened Apr 7, 2026 by Jw983cam Loading…
1
fix: enable speculative decoding for hybrid models (Qwen3.5, fixes #846)
#1111 opened Apr 5, 2026 by alexlee2046 Loading…
1
2
perf: reduce peak memory during model quantization
#1102 opened Apr 3, 2026 by matteocelani Contributor Loading…
5 tasks done
1
feat: memory-aware auto-config + BatchQuantizedKVCache for batched quantized KV
#1101 opened Apr 3, 2026 by deceptech-packet-ninja Loading…
8 tasks done
perf: reduce peak memory when loading AutoAWQ/GPTQ models
#1098 opened Apr 2, 2026 by matteocelani Contributor Loading…
4 tasks done
1
perf: reduce GPU sync frequency in GPTQ quantization
#1094 opened Apr 2, 2026 by matteocelani Contributor Loading…
3 tasks done
1
Thread local generation stream
#1090 opened Apr 2, 2026 by angeloskath Member Loading…
2
fix(gpt_oss): handle BF16 expert weights in sanitize
#1080 opened Mar 31, 2026 by FaisalFehad Loading…
1
feat: QuantizedRotatingKVCache + KVSplit (K/V different bits)
#1074 opened Mar 30, 2026 by deceptech-packet-ninja Loading…
13 tasks done
1
3
feat: add KV cache quantization args to server
#1073 opened Mar 30, 2026 by deceptech-packet-ninja Loading…
4 tasks done
1
1
ProTip! Mix and match filters to narrow down what you’re looking for.