bhoy/tb2-context-rl-rented-fa3-overnight

TB2 Context RL Rental Run

Run: tb2-context-rl-rented-fa3-overnight

Model: Qwen/Qwen3.5-9B

Benchmark/env: terminal-bench-2-context-v030

Uploaded at: 2026-07-03T00:00:28.719622+00:00

Included:

• weights/step_60, weights/step_70, weights/step_80: exported safetensors model weights.
• checkpoints/step_60, checkpoints/step_70, checkpoints/step_80: full trainer-state checkpoints.
• configs/: generated run configs.
• logs/: trainer, inference, and orchestrator logs.

The live run was stopped after step 84 to preserve the last three retained checkpoints before rotation.