-
Notifications
You must be signed in to change notification settings - Fork 981
Pull requests: ml-explore/mlx-lm
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Add M-RoPE support so vision embeddings keep their grid positions
#1768
opened Aug 21, 2026 by
lpalbou
Contributor
Loading…
Fix batch cache extract() leaking pending right padding
#1746
opened Aug 17, 2026 by
jstray
Loading…
2
Fix FunctionGemma parser for escaped braces and multiple calls
#1628
opened Jul 28, 2026 by
ask-q-ai
Loading…
[kimi_linear] Fix KDA q/k normalization eps to match reference l2norm semantics
#1624
opened Jul 27, 2026 by
avlp12
Loading…
FalconH1: apply key_multiplier to k_proj (silently dropped via a dead key_proj branch)
#1622
opened Jul 26, 2026 by
rcfa
Loading…
Fix LFM2.5 model loading: make block_ff_dim optional, fall back to intermediate_size
#1621
opened Jul 26, 2026 by
alexzhu0
Loading…
Fix rotated flag round-trip in BatchRotatingKVCache.meta_state
#1619
opened Jul 26, 2026 by
mabaeyens
Loading…
fix(granitemoehybrid): accept nested rope_parameters and tied lm_head weight
#1605
opened Jul 22, 2026 by
lkrapf
Loading…
Add Nanbeige (Nanbeige4.2 looped transformer) model support
#1597
opened Jul 21, 2026 by
MercuriusDream
Loading…
Add packed gated delta kernel, bitwise-pinned by an explicit-tree comparator
#1559
opened Jul 13, 2026 by
wyanzhao
Loading…
5 tasks done
Fix quantized KV cache on absorbed-MLA models (deepseek_v3)
#1558
opened Jul 13, 2026 by
katlun-lgtm
Loading…
3 of 4 tasks
Fix deepseek_v32 Indexer evicting attention sinks from sparse top-k
#1552
opened Jul 11, 2026 by
robertlangdonn
Contributor
Loading…
5 tasks done
Fail fast when the generation thread stops
#1514
opened Jul 10, 2026 by
dogukanveziroglu
Contributor
Loading…
Previous Next
ProTip!
Mix and match filters to narrow down what you’re looking for.