Skip to content

Pull requests: ml-explore/mlx-lm

New pull request New
172 Open 646 Closed
172 Open 646 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

Warn when speculative decoding may hurt throughput for MoE models
#1313 opened May 26, 2026 by Shylin26 Loading…
1
Expose tunable Metal memory limits in the trainer
#1312 opened May 26, 2026 by kru2710shna Loading…
1
Fix Qwen3_5 mixed-bit per-tensor quantization keys not remapped through sanitize
#1311 opened May 25, 2026 by ivaniguarans Loading…
1
Fix qwen3_coder tool parser crash on unparseable parameter values
#1310 opened May 25, 2026 by seongyun1104 Loading…
1
Adding --kv_bits as parameter.
#1309 opened May 24, 2026 by Wolfbane1 Loading…
feat(models): warn when MTP weights are discarded at load
#1306 opened May 24, 2026 by kru2710shna Loading…
4 tasks done
1
1
Fix _rstrip_until ValueError when until list is empty
#1305 opened May 23, 2026 by hadoobi Loading…
Add Gemma4 shared-KV sanitize regression
#1302 opened May 22, 2026 by rafaelescrich • Draft
Fix server XTC: accept int params and flatten special_tokens list
#1301 opened May 22, 2026 by realyxl Loading…
1
Fix enable_thinking TypeError in deepseek_v32 chat template
#1300 opened May 22, 2026 by ivaniguarans Loading…
1
generate: cache active samplers/processors in GenerationBatch hot path
#1299 opened May 22, 2026 by erayack Loading…
fix: add sanitize method to Granite model for tied embeddings
#1298 opened May 22, 2026 by SahilChachra Loading…
Add Prompt Lookup Decoding (ngram-simple) and Rolling-Hash Speculative Memory (ngram-mod)
#1297 opened May 22, 2026 by mayank2130 Loading…
1
1
Log detected tool parser on server model load
#1295 opened May 21, 2026 by robertlangdonn Loading…
1
Add Cohere2 MoE (Command A+) model support
#1294 opened May 21, 2026 by eauchs Loading…
2 of 5 tasks
11
Fix KeyError: 'name' in qwen3_coder tool parser
#1289 opened May 19, 2026 by DShickle Loading…
BatchGenerator: opt-in prefer_prefill_when_pending scheduler
#1288 opened May 19, 2026 by benjamin-levin Loading…
Fix tokenizer test failure
#1287 opened May 19, 2026 by zcbenz Collaborator Loading…
[mlx_lm] Expose 'strict' parameter in load() function
#1284 opened May 18, 2026 by zyguy Loading…
Add per-request prompt cache files to server
#1283 opened May 18, 2026 by Quiet-Node-io Loading…
Fix nemotron_h MoEGate breaking load with per-path quantization
#1282 opened May 18, 2026 by YBJ0000 Loading…
1
Add timings to server responses
#1279 opened May 16, 2026 by spicyneuron Contributor Loading…
Restrict think-state scan to assistant prefill tail
#1277 opened May 15, 2026 by eilidhmae Loading…
Add Gemma 4 assistant (MTP drafter) model class
#1276 opened May 14, 2026 by broomva Loading…
fix: make generation_stream per-thread to fix server crash on worker threads
#1275 opened May 14, 2026 by nish2292 Loading…
4 tasks done
1
1
ProTip! Follow long discussions with comments:>50.