Alex4210987/qwen-tilelang-infer
High-performance TileLang fused kernels for Qwen3 inference (RMSNorm, RoPE, GQA Flash Attention, SwiGLU, W4A16 GEMV) with auto-tuning on A100
CS PhD @Fudan (Fall '26). Formerly ML Frameworks Intern @AMD(@ROCm), Research Intern @ICT, CAS(@wangchenxi7 )
High-performance TileLang fused kernels for Qwen3 inference (RMSNorm, RoPE, GQA Flash Attention, SwiGLU, W4A16 GEMV) with auto-tuning on A100
Agentic Kernel Optimization for All — automated GPU kernel optimization for any kernel, any hardware, any language
Qwen3 TileLang kernel mini project for Computing and Intelligence
Standalone AscendC Sinkhorn A2 implementation from KernelSwift-FISAC
FlyDSL is the Python front‑end of the project: Flexible LaYout DSL.
Ascend TileLang adapter
Domain-specific language designed to streamline the development of high-performance GPU/CPU/Accelerators kernels
assembly oj lab & hw of hust-cse
hust cse computer network lab
hust-cse-db-lab
HIDS base on Kairos GNN and Linux Audit