The DOI matches its Crossref record.
Re-ReST: Reflection-Reinforced Self-Training for Language Agents
Open accessAbstract
Finetuning language agents with reasoningaction trajectories is effective, but obtaining these trajectories from human annotations or stronger models is costly and sometimes impractical.In this paper, we investigate the use of self-training in language agents, which can generate supervision from the agent itself, offering a promising alternative without relying on human or stronger model demonstrations.Self-training, however, requires high-quality model-generated samples, which are hard to obtain for challenging language agent tasks.To address this, we present Reflection-Reinforced Self-Training (Re-ReST), which uses a reflector to refine low-quality generated samples during self-training.The reflector takes the agent's output and feedback from an external environment (e.g., unit test results in code generation) to produce improved samples.This technique enhances the quality of inferior samples and efficiently enriches the self-training dataset with higher-quality samples.We conduct extensive experiments on open-source language agents across tasks, including multi-hop question answering, sequential decision-making, code generation, visual question answering, and text-toimage generation.The results demonstrate the effectiveness of self-training and Re-ReST in language agent tasks, with self-training improving baselines by 7.6% on HotpotQA and 28.4% on AlfWorld, and Re-ReST further boosting performance by 2.0% and 14.1%, respectively.Our studies also confirm the efficiency of using a reflector to generate high-quality samples for self-training.Moreover, we demonstrate a method to employ reflection during inference without ground-truth feedback, addressing the limitation of previous reflection work.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Re-ReST augments the STaR recipe with a reflector that repairs failed trajectories using environment feedback, converting failures into training signal instead of discarding them.
“Re-ReST augments it with a reflector that repairs failed trajectories using environment feedback, converting failures into training signal instead of discarding them [95]”
From Recursive Self-Improvement in AI: From Bounded Self-Refinement to Autonomous Research Loops (2026) · cites this paper as [95]
Citations
Cites
Still being checked.
Cited by
Sorted by citations
Evaluation and Benchmarking of LLM Agents: A Survey
Mohammadi, M., Li, Y., Lo, J., & Yip, W. (2025) · Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2 · doi:10.1145/3711896.3736570 · Cited by 74
From screens to scenes: A survey of embodied AI in healthcare
Liu, Y., Cao, X., Chen, T., Jiang, Y., You, J., Wu, M., Wang, X., Feng, M., Jin, Y., & Chen, J. (2025) · Information Fusion · doi:10.1016/j.inffus.2025.103033 · Cited by 76
LLM-A*: Large Language Model Enhanced Incremental Heuristic Search on Path Planning
Meng, S., Wang, Y., Yang, C., Peng, N., & Chang, K.-W. (2024) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2024.findings-emnlp.60 · Cited by 50
Self-citationLLM-A*: Large Language Model Enhanced Incremental Heuristic Search on Path PlanningPreprint
Meng, S., Wang, Y., Yang, C., Peng, N., & Chang, K.-W. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2407.02511 · Cited by 41
Self-citationA Survey on Autonomy-Induced Security Risks in Large Model-Based Agents
Su, H., Luo, J., Liu, C., Yang, X., Zhang, Y., Dong, Y., & Zhu, J. (2026) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2026.3688650 · Cited by 11
MatchNav: LLM-based enhanced description and instruction matching in vision-and-language navigation
Wang, C., Wei, S., & Juntong, Q. (2025) · Information Fusion · doi:10.1016/j.inffus.2025.103444 · Cited by 5
Meta-Reflection: A Feedback-Free Reflection Learning Framework
Wang, Y., Zhu, Y., XintongBao, X., Zhang, W., Dai, S., Chen, K., Li, W., Huang, G., Tang, S., & Zhuang, Y. (2025) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2025.acl-long.201 · Cited by 3
Ambiguity Resolution in Vision-and-Language Navigation with Large Language Models
Wei, S., Wang, C., & Qi, J. (2024) · IEEE International Conference on Information Technology, Big Data and Artificial Intelligence (ICIBA) · doi:10.1109/iciba62489.2024.10868019 · Cited by 1
Recursive Self-Improvement in AI: From Bounded Self-Refinement to Autonomous Research Loops
Chen, M., Wang, L., & Qu, B. (2026) · arXiv
“…with a reflector that repairs failed trajectories using environment feedback, converting failures into training signal instead of discarding them [95]; PRefLexOR recasts the loop as recursive preference optimization over intermediate reasoning steps [96]; LaTRO shows the loop can run with no…”
SIPA: a self-iterative preference alignment method for generative language models
Du, Y., Wang, B., Hou, Y., & Han, H. (2026) · Applied Intelligence · doi:10.1007/s10489-025-07079-9
LLM-GA: Large Language Model-Enhanced Genetic Algorithm for Path Optimization
Sun, M., Hao, G., Wang, X., Feng, X., Zhu, Y., & Ren, S. (2025) · Asian Conference on Artificial Intelligence Technology (ACAIT) · doi:10.1109/acait67930.2025.11522566
Reflection-Augmented Strategy Generation via In-Game Triggering in Large Language Models
Ni, Y., Zhu, Z., 龚晨枫, Qi, R., & Huang, K. (2026) · International Conference on Big Data and Information Analytics (BigDIA) · doi:10.1109/bigdia71496.2026.11684683
Ren, Y., Yu, Y., Liao, L., Shang, Y., Wang, Y., Lu, K., Guo, R., Cheng, N., & Li, S. (2026) · Information Processing & Management · doi:10.1016/j.ipm.2026.105276
Metadata
- Authors
- Dou, Z.-Y., Yang, C.-F., Wu, X., Chang, K.-W., & Peng, N.
- Affiliations
- University of California, Los Angeles (US)
- Venue
- Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing
- Published
- 2024
- Pages
- 15394-15411
- OpenAlex
- W4404782171
- Cited by
- 6
- References
- 0
- Citations by year
- 2026: 2 · 2025: 3 · 2024: 1
- FWCI
- 1.73
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- dx.doi.org
- Licence
- cc-by
- Locations
- Conference on Empirical Methods in Natural Language Processing (EMNLP) (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion, cc-by)
- arXiv (Cornell University) (cc-by)
- Funders
- Cisco Systems
- Defense Advanced Research Projects Agency
- Office of Naval Research (N00014-23-1-2780, N00014)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
- Physical Sciences › Computer Science › Artificial Intelligence › Reinforcement Learning in Robotics
- Keywords
- language agents, self-training, reflection, reflector, fine-tuning, code generation, multi-hop question answering, sequential decision making, visual question answering, text-to-image generation
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
