lablup/mlxcel

feat: need a logo 15 days ago

help wanted status:backlog type:enhancement priority:medium area:docs

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

docs: fix garbled doc comments left behind by issue-reference scrubbing 20 days ago

good first issue status:done type:docs priority:low

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

docs: correct the MLXCEL_DEVICE row in installation.md (wrong default, missing metal value) 20 days ago

good first issue status:review type:docs priority:low area:docs

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

fix(bench): speculative_bench panics on a user-supplied --target path with no file name 21 days ago

good first issue status:ready type:bug priority:low area:cli area:benchmark

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

fix(bench): format_bytes in the KV-transfer benchmark divides by 1024 but prints decimal GB/MB/KB 23 days ago

good first issue status:in-progress type:bug priority:low area:benchmark

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

docs: fix the docs/README.md index intro and entry separator inconsistency about 1 month ago

good first issue status:ready type:docs priority:low area:docs

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

refactor(cli): deduplicate eleven identical "Failed to load image" closures into one open_image helper about 1 month ago

good first issue status:ready type:refactor priority:low area:cli

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

chore(cli): unify error-message capitalization across CLI commands about 1 month ago

good first issue status:ready type:chore priority:low area:cli

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

docs: replace the opaque (A4:) and (B11:) task codes in rustdoc with the (issue #NNN) convention about 1 month ago

good first issue status:ready type:docs priority:low

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

docs(bench): clean scrub artifacts in benchmark_results (broken link, headless sentence, non-rendering bold, dangling prefixes) about 1 month ago

good first issue status:ready type:docs priority:low area:docs area:benchmark

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

docs: sync the PR template's clippy and test checklist commands with CONTRIBUTING.md about 1 month ago

good first issue status:ready type:docs priority:low area:docs

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

refactor(server): remove or wire up the unused build_failed_response helper in responses_translator.rs about 1 month ago

good first issue status:ready type:refactor priority:low area:inference

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

refactor(loading): remove the dead is_special_weight_model_type delegate wrapper about 1 month ago

good first issue status:ready type:refactor priority:low area:models

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

docs(python): quote the pip extras argument so the install commands work under zsh about 1 month ago

good first issue status:ready type:docs priority:low area:docs

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

docs: update SECURITY.md supported versions and the bug-report version placeholder to the 0.5.x line about 1 month ago

good first issue status:ready type:docs priority:low area:docs

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

bug: Gemma 4 server reasoning_content retains leading 'thought' prefix from the channel marker about 1 month ago

good first issue status:done type:bug priority:low

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

fix: InvalidConfig labels non-LLaVA loader errors as invalid LLaVA config about 2 months ago

good first issue status:ready type:bug priority:low area:models

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

fix(tests): fused MoE GeGLU parity threshold is tighter than the gather_qmm reference's own jitter about 2 months ago

good first issue status:ready type:test priority:medium area:core

lablup/mlxcel

134

High-performance LLM/VLM inference runtime and server for Apple Silicon / CUDA devices

Rust

Open Issues Need Help