dice-cuda-generation

【开发与工程】Train diffusion language models to generate optimized CUDA kernels using bi-phase reinforcement learning. First phase masks and regenerates core kernel logic with provided scaffolding to prevent PyTorch shortcuts. Second phase enables end-to-end generation. Leverage diffusion's global context awareness for non-sequential code generation.

独立 GitHub 快照 · 自动分类 · 以原作者文档为准