Language agents improve long-horizon performance by generating and refining their own internal reward signals that guide actions at inference and provide denser supervision during training.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Co-Evolution of Policy and Internal Reward for Language Agents
Language agents improve long-horizon performance by generating and refining their own internal reward signals that guide actions at inference and provide denser supervision during training.