Show HN: I RL-trained an agent that trains models with RL (for ~$1.3k)github.com·107 pts·Danau5tin·49
Scaling Coding-Agent RL to 32x H100s. 160% Improvement on Stanford's TBenchgithub.com·2 pts·Danau5tin·1
Show HN: Multi-Agent-Coder Is #12 on Stanford's TBench. Beats Claude Codegithub.com·5 pts·Danau5tin·1
My weekend project accidentally beat Claude Code – #12 on Stanford's TBenchgithub.com·2 pts·Danau5tin·2
Show HN: Terminal-Bench-RL: Training long-horizon terminal agents with RLgithub.com·125 pts·Danau5tin·12