lablup/mlxcel
View on GitHubHigh-performance LLM, VLM, embedding, reranking, and audio inference for Apple Silicon and NVIDIA CUDA systems.
- Stars
- 467
- Forks
- 54
- Open beginner issues
- 0
- Indexed issues
- 152
- Avg merge
- 4h 25m
- Merged PRs (30d)
- 310
- Dominant language
- Rust
- License
- Apache-2.0
- Last GitHub push
- Sep 15, 2026
- Latest indexed
- Sep 15, 2026
- Contributing guide
- Contributing guide
- Code of conduct
- Code of conduct
- Beginner labels
- help wanted
-
priority:low status:backlog type:chore
-
priority:medium status:ready type:enhancement
-
area:core area:inference platform:windows priority:medium status:investigation type:enhancement
-
feat: need a logo Openarea:docs help wanted priority:medium status:backlog type:enhancement
-
priority:medium status:ready type:enhancement
-
area:architecture area:core area:inference priority:high status:in-progress type:enhancement
-
feat: multi-GPU tensor-parallel runtime with per-rank device placement and cross-GPU collectives Openarea:architecture area:inference priority:high status:ready type:enhancement
lablup/mlxcel#488 · 0 comments · 0 reactions · 0 assignees ·
-
area:inference status:in-progress type:enhancement
-
area:models priority:low status:investigation type:enhancement
-
area:models priority:low status:investigation type:enhancement
lablup/mlxcel#548 · 0 comments · 0 reactions · 0 assignees ·
-
epic: OpenXLA SSM / hybrid / recurrent architecture support (Mamba2, Nemotron-H, Jamba, and more) Openarea:architecture area:inference priority:low status:ready type:enhancement
-
epic: OpenXLA multimodal / VLM architecture support (Qwen2-VL, Gemma3n, Phi4MM, Molmo, and more) Openarea:architecture area:inference priority:low status:blocked type:enhancement
-
area:inference priority:low status:blocked type:performance
-
area:inference priority:low status:ready type:enhancement
lablup/mlxcel#603 · 0 comments · 0 reactions · 0 assignees ·
-
arch:hybrid area:models modelsize:xlarge modeltype:text priority:medium status:blocked type:enhancement
lablup/mlxcel#842 · 0 comments · 0 reactions · 0 assignees ·
-
arch:dense area:models modeltype:text priority:medium status:blocked type:enhancement
-
area:inference area:models priority:low status:in-progress type:enhancement
lablup/mlxcel#850 · 2 comments · 0 reactions · 0 assignees ·
-
area:benchmark area:inference priority:medium status:ready type:test
lablup/mlxcel#851 · 0 comments · 0 reactions · 0 assignees ·
-
area:inference area:models priority:low status:review type:enhancement
lablup/mlxcel#866 · 0 comments · 0 reactions · 0 assignees ·
-
area:inference area:models priority:low status:review type:enhancement
lablup/mlxcel#868 · 0 comments · 0 reactions · 0 assignees ·
-
area:inference area:models priority:low status:review type:enhancement
lablup/mlxcel#869 · 0 comments · 0 reactions · 0 assignees ·
-
area:inference area:models priority:low status:review type:enhancement
lablup/mlxcel#870 · 0 comments · 0 reactions · 0 assignees ·
-
area:inference area:models priority:low status:review type:enhancement
lablup/mlxcel#872 · 0 comments · 0 reactions · 0 assignees ·
-
area:architecture area:inference area:models priority:low status:blocked type:enhancement
lablup/mlxcel#878 · 2 comments · 0 reactions · 0 assignees ·
-
area:architecture area:core area:inference priority:high status:blocked type:enhancement
lablup/mlxcel#932 · 11 comments · 0 reactions · 0 assignees ·
-
fix(core): f32 matmul and single-query SDPA lose ~fp16 precision on M5, breaking four numeric tests Openarea:core area:inference platform:macos priority:high status:ready type:bug
lablup/mlxcel#1065 · 4 comments · 0 reactions · 0 assignees ·
-
priority:low status:ready type:bug
lablup/mlxcel#1131 · 0 comments · 0 reactions · 0 assignees ·
-
priority:low status:ready type:chore
lablup/mlxcel#1140 · 0 comments · 0 reactions · 0 assignees ·
-
priority:low status:ready type:chore
lablup/mlxcel#1141 · 0 comments · 0 reactions · 0 assignees ·
-
area:benchmark area:inference platform:macos priority:medium status:ready type:performance
lablup/mlxcel#1217 · 1 comment · 0 reactions · 0 assignees ·
-
area:inference platform:macos priority:high status:ready type:bug
lablup/mlxcel#1279 · 1 comment · 0 reactions · 0 assignees ·
-
area:inference platform:macos priority:medium status:ready type:performance
lablup/mlxcel#1289 · 0 comments · 0 reactions · 0 assignees ·
-
arch:dense area:docs area:inference area:models modelsize:small modeltype:text priority:low status:ready type:enhancement
lablup/mlxcel#1363 · 0 comments · 0 reactions · 0 assignees ·
-
arch:dense area:cli area:docs area:inference area:models modelsize:small modeltype:vlm priority:medium status:ready type:enhancement
lablup/mlxcel#1365 · 0 comments · 0 reactions · 0 assignees ·
-
area:inference area:models modeltype:text modeltype:vlm priority:low status:ready type:enhancement
lablup/mlxcel#1367 · 0 comments · 0 reactions · 0 assignees ·
-
arch:dense area:docs area:inference area:models modelsize:small modeltype:text priority:low status:ready type:enhancement
lablup/mlxcel#1368 · 0 comments · 0 reactions · 0 assignees ·
-
arch:dense area:cli area:docs area:inference area:models modelsize:small modeltype:vlm priority:medium status:ready type:enhancement
lablup/mlxcel#1369 · 0 comments · 0 reactions · 0 assignees ·
-
arch:dense area:core area:docs area:inference area:models modelsize:small modeltype:text priority:medium status:ready type:enhancement
lablup/mlxcel#1370 · 0 comments · 0 reactions · 0 assignees ·
-
arch:hybrid area:cli area:docs area:inference area:models modelsize:small modeltype:audio priority:medium status:ready type:enhancement
lablup/mlxcel#1372 · 0 comments · 0 reactions · 0 assignees ·
-
area:cli area:core area:inference modeltype:text priority:low status:blocked type:enhancement
lablup/mlxcel#1373 · 0 comments · 0 reactions · 0 assignees ·
-
arch:hybrid area:cli area:docs area:inference area:models modelsize:small modeltype:audio priority:medium status:ready type:enhancement
lablup/mlxcel#1374 · 0 comments · 0 reactions · 0 assignees ·
-
area:cli area:docs area:inference modeltype:audio priority:medium status:blocked type:enhancement
lablup/mlxcel#1376 · 0 comments · 0 reactions · 0 assignees ·
-
arch:hybrid area:docs area:inference area:models modelsize:small modeltype:audio priority:medium status:blocked type:enhancement
lablup/mlxcel#1378 · 0 comments · 0 reactions · 0 assignees ·
-
feat(plamo2vl): port PLaMo 2.1 VL (SigLIP tiles + RMSNorm/GELU adapter + PLaMo 2 hybrid decoder) Openarch:hybrid area:cli area:docs area:inference area:models modelsize:small modeltype:vlm priority:low status:ready type:enhancement
lablup/mlxcel#1380 · 0 comments · 0 reactions · 0 assignees ·
-
area:core platform:linux priority:medium status:ready type:bug
lablup/mlxcel#1422 · 0 comments · 0 reactions · 0 assignees ·
-
refactor(embeddings): hoist family-local weight-key handling into the shared embedding sanitizer Openarea:models priority:low status:ready type:refactor
lablup/mlxcel#1424 · 0 comments · 0 reactions · 0 assignees ·
-
area:inference priority:low status:ready type:enhancement
lablup/mlxcel#1426 · 0 comments · 0 reactions · 0 assignees ·
-
area:architecture priority:low status:backlog type:security
lablup/mlxcel#1457 · 0 comments · 0 reactions · 0 assignees ·
-
area:core platform:linux priority:medium status:in-progress type:performance
-
area:core platform:linux priority:high status:investigation type:performance
lablup/mlxcel#1543 · 3 comments · 0 reactions · 0 assignees ·