automegakernel
An agent harness that compiles a model into one provably-correct, self-retargeting CUDA megakernel and self-tunes it past cuBLAS at batch-1 LLM decode, paper: https://arxiv.org/abs/2606.09682
Take a look →OPEN SOURCE, IN FOCUS
retro-harness
wbopan/retro-harness
RHO: Retrospective Harness Optimization — improving LLM agents from unlabeled past trajectories (arXiv:2606.05922).
Saved projects stay in this browser. No account needed.
TAKE A CLOSER LOOK
Start with the source.
The repository is the place to check installation steps, supported platforms, the license, and current maintenance before choosing a tool.
Read the documentation on GitHub ↗Keep exploring
Find projects in the same areas, or search the archive for a different approach.