The authors replayed textual trajectories from a frontier model inside each target model, without accessing the source model’s logits or hidden states.
Text traces can steer frozen agents without additional training
REVEAL converts successful and failed frontier-model trajectories into activation-level controls tailored to smaller open-source agents.
Chinese Tech
Yuetian Du · Meng Wang · Yucheng Wang · Zhangzhe Chen · Wei Wang · Boyu Yang · +6 more
Zhejiang University · South China University of Technology · Alibaba Group · Fudan University · City University of Hong Kong
Research Digest··2 min read
Thread:Agent Self-Improvement
Du and colleagues present a training-free method for transferring experience from frontier-model interaction traces to frozen target models.
Why this paper
From Alibaba Group and 4 others · Part of Agent Self-Improvement, now 52 papers
In one line
REVEAL steers frozen open-source agents using textual frontier model trajectories, improving task completion without training or internal access.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§