HN Explorer

Danau5tin

↗ HN profile [ 50.2 ms ] full profile
Karma
0
Created
()
Submissions
0
  1. Show HN: I RL-trained an agent that trains models with RL (for –$1.3k) (github.com)
    4 points 35
  2. Scaling Coding-Agent RL to 32x H100s. 160% Improvement on Stanford's TBench (github.com)
    2 points 1
  3. Show HN: Multi-Agent-Coder Is #12 on Stanford's TBench. Beats Claude Code (github.com)

    This weekend I built a multi-agent coding system which, quite unexpectedly, beat Claude Code on Stanford's Terminal Bench! The architecture is straightforward, consisting of an orchestrator agent that deploys explorer &…

    5 points 1
  4. My weekend project accidentally beat Claude Code – #12 on Stanford's TBench (github.com)
    2 points 2
  5. Show HN: Terminal-Bench-RL: Training long-horizon terminal agents with RL (github.com)

    After training calculator agent via RL, I really wanted to go bigger! So I built RL infrastructure for training long-horizon terminal/coding agents that scales from 2x A100s to 32x H100s (~$1M worth of compute!) Without…

    125 points 11