ShangYixin
Community@qinghuan-6
I'm ShangYixin,a student in Wuhan University of Technology.I want to contribute to the open source community.
Agent Skills by ShangYixin
Showing 19 vetted skills indexed across 1 GitHub repositories.
QingHuan-6sglang-diffusion-modelopt-quant
Quantize diffusion DiT models with ModelOpt to FP8 or NVFP4 checkpoints for SGLang Diffusion.
QingHuan-6sglang-diffusion-benchmark-profile
Measure diffusion denoise latency and collect performance traces for SGLang.
QingHuan-6sglang-diffusion-add-model
Automate integration of diffusion models into SGLang with configuration and validation.
QingHuan-6sglang-diffusion-ako4all-kernel
Automate AKO4ALL kernel tuning for SGLang diffusion kernels with validation.
QingHuan-6sglang-diffusion-performance
Identify fastest SGLang Diffusion flags for a model, GPU, and VRAM budget.
QingHuan-6clean-startup-log
Filter non-actionable warnings from SGLang server startup logs.
QingHuan-6write-sglang-test
Create CI/UT tests for SGLang projects using CustomTestCase and registered placement.
QingHuan-6ci-workflow-guide
Automate CI workflow orchestration for SGLang project testing.
QingHuan-6add-jit-kernel
Automate adding, validating, and benchmarking JIT CUDA kernels for SGLang.
QingHuan-6debug-cuda-crash
Diagnose CUDA crashes in SGLang by exposing kernel API call traces and tensor metadata.
QingHuan-6llm-serving-auto-benchmark
Benchmark SGLang, vLLM, and TensorRT-LLM serving to produce Markdown/CSV summaries.
QingHuan-6sglang-prod-incident-triage
Convert live SGLang serving symptoms into reproducible replays and structured next-step guidance.
QingHuan-6generate-profile
Profile SGLang server performance by capturing Chrome-compatible traces.
QingHuan-6add-sgl-kernel
Implement AOT CUDA/C++ kernels in sgl-kernel with tests and benchmarks.
QingHuan-6llm-torch-profiler-analysis
Triage torch-profiler traces into kernel, overlap, and fuse tables across LLM frameworks.
QingHuan-6sglang-bisect-ci-regression
Extract failure signatures and bisect SGLang CI regressions across main-branch commits.
QingHuan-6sglang-sota-performance
Compare SGLang, vLLM, and TensorRT-LLM deployment commands and generate benchmark reports.
QingHuan-6debug-distributed-hang
Diagnose distributed SGLang hangs across TP/PP/DP/EP with per-rank logging and binary-search root-cause analysis.
QingHuan-6mechanical-refactor-verify
Verifies that a scripted transform reproduces a mechanical refactor with an empty diff.