The authors propose RECAP (REdundancy-aware Credit Assignment via Propagation), which combines two signals: structural responsibility, measured by propagating credit backward through an LLM-annotated semantic dependency graph from the final answer node, and step efficacy, measured by the change in gold-answer log-likelihood as each step is added.
Credit assignment that prunes redundant steps improves reasoning accuracy and efficiency
The authors' RECAP method reshapes GRPO advantages into step-specific updates using semantic dependency graphs and answer-directed progress, shortening reasoning traces without hurting accuracy.
Big Tech
Yuqing Zhou · Hong Wang · Manqing Mao · Zhuoer Wang · Samson Koelle · Jie Yuan · +5 more
George Mason University · Amazon, Inc
Research Digest··2 min read
The authors introduce RECAP, a redundancy-aware credit assignment method that assigns step-level rewards based on both a step's downstream role in the reasoning structure and its progress toward the correct answer.
Why this paper
From Amazon, Inc and George Mason University · Part of Credit Assignment in Agentic RL, now 30 papers
In one line
RECAP assigns credit based on downstream dependency and correct-answer progress, improving accuracy and reducing reasoning tokens in large models.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ·No stated limitations found
- ✓Reports numbers on named benchmarks
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§