TritonForge

Public

🔥 LLM-powered GPU kernel synthesis: Train models to convert PyTorch ops into optimized Triton kernels via SFT+RL. Multi-turn compilation feedback, cross-platform NVIDIA/AMD, Kernelbook + KernelBench

Updated Jan 31, 2026

Created Sep 9, 2025

116 stars

2 forks

2 watchers

2 open issues

Languages

Codebase composition by bytes

Top Contributors

113 commits

102 commits

96 commits

91 commits

15 commits

12 commits

11 commits

7 commits

7 commits

5 commits

Recent Stargazers

5 months ago

5 months ago

5 months ago

5 months ago

5 months ago

5 months ago

5 months ago

5 months ago

5 months ago

5 months ago

Issues & Activity

Open vs closed issues and recent activity peak.

Open Issues 2Closed Issues 1

Open Issues3

Activity Peak

Most recent day with highest activityJan 31, 2026 • 1 events

GitHub Analytics

TritonForge