$ git log --author="hnyls2002" --all --oneline

On the record

Liangsheng Yin

@hnyls2002 ↗ first byline July 2026

Every morning, RepoJournal's newsroom reads what shipped across the open-source projects developers depend on — and writes the wire. Your commits kept making the news. This is your clippings file: the stories where your work was the story.

stories filed

daily wires

$ grep -rn "author: hnyls2002" wires/ | sort -r

Local LLMs

wire 2026-10-30

Faster Metal MXFP4 decode, a GLM-4.7 streaming parser stall fixed

[Fix] disaggregation: Keep output ids during a PD rebootstrap instead of popping and replaying the last one sgl-project/sglang

Local LLMs

wire 2026-10-13

Metal MoE fix for duplicate ids, sglang stops crashing on retracted DeepSeek-V4

[PD] Recompute retracted DeepSeek-V4 requests on prefill instead of crashing decode sgl-project/sglang

Local LLMs

wire 2026-10-10

vLLM's unauthenticated DoS surface, and a 1.9x SYCL MXFP4 lift

[Scheduler] Fix the deterministic-inference hang when a prefill chunk is shorter than its alignment sgl-project/sglang

Local LLMs

wire 2026-10-09

vLLM drops compile-based sequence parallelism, llama.cpp fixes slot checkpoints and a silent CUDA fallback

[MemCache] Delete `ChunkCache` sgl-project/sglang

Local LLMs

wire 2026-10-04

OpenVINO 2026.4.1 lands in llama.cpp, vLLM frees CUDA graph memory on sleep

[mem_cache] Run mamba models on `UnifiedRadixCache` when the radix cache is disabled sgl-project/sglang

Local LLMs

wire 2026-09-23

Ollama folds structured output into one pass, llama.cpp binds multiple addresses

[HiCache] Remove the unused HiRadixCache sgl-project/sglang

Local LLMs

wire 2026-09-08

SGLang fixes TP hangs and GLM-5.2 routing

Revert "[kernel] add fused silu mul quant fp8" sgl-project/sglang

Local LLMs

wire 2026-08-27

vLLM 0.28.0 LANDED WITH 584 COMMITS AND A KIMI-K3 PERFORMANCE PUSH

[Feature] Beam search support sgl-project/sglang

Local LLMs

wire 2026-08-20

OLLAMA FIXES DEADLOCK THAT HUNG CHAT AND GENERATE

Revert "[Feature] Add DeepEPv2 (ElasticBuffer) MoE A2A backend" sgl-project/sglang

Local LLMs

wire 2026-08-10

OLLAMA SHIPS VISION SUPPORT ACROSS MLX AND QWEN3.5 AS LLAMA.CPP FIXES DISPATCH BUGS

[DSV4] Fix silent KV corruption when speculative draft tokens > 4 sgl-project/sglang

Local LLMs

wire 2026-08-09

OLLAMA CUTS MODEL CONFIG BLOAT 100KB+ PER MODEL, LLAMA.CPP SHIPS TOOL ISOLATION AND CUDA FUSION GAINS

[Scheduler] Unify WAR read-done gating behind shared-read boundary declarations sgl-project/sglang
[CI] Trim redundant nightly test registrations sgl-project/sglang

Local LLMs

wire 2026-08-05

OLLAMA FIXES RACE CONDITIONS AS LLAMA.CPP SHIPS BREAKING TTS CHANGES

[Fix] Resolve VLM test image placeholders from the model's own chat template sgl-project/sglang

Local LLMs

wire 2026-08-03

DEEPSEEK V4 INFERENCE GETS 50% SPEEDUP; LLAMA.CPP AND VLLM BOTH SHIPPING OPTIMIZED PATHS

[Spec] Support sampling in the DSPARK graph-folded draft proposal sgl-project/sglang
[Perf] Broadcast single-image DP vision embedding instead of pad-to-max all-gather sgl-project/sglang

Local LLMs

wire 2026-07-29

OLLAMA RIPS OUT IMAGE GENERATION, LLAMA.CPP SHIPS SPECULATIVE DECODING

[Refactor] Remove dead allocator `backup_state` / `restore_state` (#32709) sgl-project/sglang