AgentPTB run state
Updated: 2026-08-19 09:57 UTC Hours left: ~2.4 Cell: grok / Qwen3.5-9B-Base / 4x B200
Submit
- Weights:
outputs/soup-85-s80/weights - Fallback:
outputs/rl-r2e4/weights/step_60 - Harness:
pi+ skills - Eval: TB2 192/0.2, SWE 512/0.2, :8600
- Combined: TB2 5/102, SWE 10/128
Status
Serve healthy :8600 soup-85-s80. Submit unchanged.