Agent Skills by tuliang1024
Showing 5 vetted skills indexed across 1 GitHub repositories.
tuliang1024kvcache-optimization
Optimize KVCache handling for LLM inference to reduce memory usage and improve throughput.
Community
Advanced
tuliang1024model-optimize
Orchestrate staged KVCache, fusion, and graph-mode optimizations for NPU inference.
Community
Advanced
tuliang1024kvcache-fa-precision-debug
Diagnose precision issues from KVCache writes and Flash Attention fusion operators.
Community
Advanced
tuliang1024torch-npu-fusion-optimizer
Analyze PyTorch model code to generate torch_npu fusion-operator replacement blueprints.
Community
Advanced
tuliang1024graph-mode-adaptation
Adapt models to NPU graph mode acceleration with torch.compile.
Community
Advanced