Paper
Retrospective In-Context Learning for Temporal Credit Assignment with Large Language Models
Authors
Wen-Tse Chen, Jiayu Chen, Fahim Tajwar, Hao Zhu, Xintong Duan, Ruslan Salakhutdinov, Jeff Schneider
Abstract
Learning from self-sampled data and sparse environmental feedback remains a fundamental challenge in training self-evolving agents. Temporal credit assignment mitigates this issue by transforming sparse feedback into dense supervision signals. However, previous approaches typically depend on learning task-specific value functions for credit assignment, which suffer from poor sample efficiency and limited generalization. In this work, we propose to leverage pretrained knowledge from large language models (LLMs) to transform sparse rewards into dense training signals (i.e., the advantage function) through retrospective in-context learning (RICL). We further propose an online learning framework, RICOL, which iteratively refines the policy based on the credit assignment results from RICL. We empirically demonstrate that RICL can accurately estimate the advantage function with limited samples and effectively identify critical states in the environment for temporal credit assignment. Extended evaluation on four BabyAI scenarios show that RICOL achieves comparable convergent performance with traditional online RL algorithms with significantly higher sample efficiency. Our findings highlight the potential of leveraging LLMs for temporal credit assignment, paving the way for more sample-efficient and generalizable RL paradigms.
Metadata
Related papers
Vibe Coding XR: Accelerating AI + XR Prototyping with XR Blocks and Gemini
Ruofei Du, Benjamin Hersh, David Li, Nels Numan, Xun Qian, Yanhe Chen, Zhongy... • 2026-03-25
Comparing Developer and LLM Biases in Code Evaluation
Aditya Mittal, Ryan Shar, Zichu Wu, Shyam Agarwal, Tongshuang Wu, Chris Donah... • 2026-03-25
The Stochastic Gap: A Markovian Framework for Pre-Deployment Reliability and Oversight-Cost Auditing in Agentic Artificial Intelligence
Biplab Pal, Santanu Bhattacharya • 2026-03-25
Retrieval Improvements Do Not Guarantee Better Answers: A Study of RAG for AI Policy QA
Saahil Mathur, Ryan David Rittner, Vedant Ajit Thakur, Daniel Stuart Schiff, ... • 2026-03-25
MARCH: Multi-Agent Reinforced Self-Check for LLM Hallucination
Zhuo Li, Yupeng Zhang, Pengyu Cheng, Jiajun Song, Mengyu Zhou, Hao Li, Shujie... • 2026-03-25
Raw Data (Debug)
{
"raw_xml": "<entry>\n <id>http://arxiv.org/abs/2602.17497v1</id>\n <title>Retrospective In-Context Learning for Temporal Credit Assignment with Large Language Models</title>\n <updated>2026-02-19T16:13:28Z</updated>\n <link href='https://arxiv.org/abs/2602.17497v1' rel='alternate' type='text/html'/>\n <link href='https://arxiv.org/pdf/2602.17497v1' rel='related' title='pdf' type='application/pdf'/>\n <summary>Learning from self-sampled data and sparse environmental feedback remains a fundamental challenge in training self-evolving agents. Temporal credit assignment mitigates this issue by transforming sparse feedback into dense supervision signals. However, previous approaches typically depend on learning task-specific value functions for credit assignment, which suffer from poor sample efficiency and limited generalization. In this work, we propose to leverage pretrained knowledge from large language models (LLMs) to transform sparse rewards into dense training signals (i.e., the advantage function) through retrospective in-context learning (RICL). We further propose an online learning framework, RICOL, which iteratively refines the policy based on the credit assignment results from RICL. We empirically demonstrate that RICL can accurately estimate the advantage function with limited samples and effectively identify critical states in the environment for temporal credit assignment. Extended evaluation on four BabyAI scenarios show that RICOL achieves comparable convergent performance with traditional online RL algorithms with significantly higher sample efficiency. Our findings highlight the potential of leveraging LLMs for temporal credit assignment, paving the way for more sample-efficient and generalizable RL paradigms.</summary>\n <category scheme='http://arxiv.org/schemas/atom' term='cs.LG'/>\n <published>2026-02-19T16:13:28Z</published>\n <arxiv:comment>Accepted to NeurIPS 2025</arxiv:comment>\n <arxiv:primary_category term='cs.LG'/>\n <author>\n <name>Wen-Tse Chen</name>\n </author>\n <author>\n <name>Jiayu Chen</name>\n </author>\n <author>\n <name>Fahim Tajwar</name>\n </author>\n <author>\n <name>Hao Zhu</name>\n </author>\n <author>\n <name>Xintong Duan</name>\n </author>\n <author>\n <name>Ruslan Salakhutdinov</name>\n </author>\n <author>\n <name>Jeff Schneider</name>\n </author>\n </entry>"
}