Terminal-Bench 2.0

Mentions · 7d0
Mentions · 30d0
All mentions1
First seenSep 17, 2026
Last mentionedSep 17, 2026

Timeline

1 articles · newest first

Don't Mask the Environment: Observation Supervision Changes How Agents Explore Under RL

ActObs supervises observation tokens during SFT, improving GRPO-trained agent pass@k on Terminal-Bench 2.0 and cross-domain code editing for Qwen3 models.

arXiv cs.AI / cs.LG / cs.CLSep 17, 17:11 UTC · Sep 17, 2026AI research1 · 1 read

Appears with

Qwenvendor · 1
Qwen3-8Bai model · 1
Qwen3-4Bai model · 1
Qwen3-8Bproduct · 1
Qwen3-4Bproduct · 1

Entities are extracted by the model from each article. First seen Sep 17, 2026, last mentioned Sep 17, 2026. Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.