Research map / Multi-agent systems and games / LLM agents
Reflexion: language agents with verbal reinforcement learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add Reflexion verbal self-reflection loop | GPT-4 · HumanEval (PY) | pass@1: 80.1 -> 91.0 | Helped |
| add Reflexion | GPT-4 · MBPP (PY) | pass@1: 80.1 -> 77.1 | Hurt |
| add Reflexion | GPT-4 · Leetcode Hard (PY) | pass@1: 7.5 -> 15.0 | Helped |
| add Reflexion | ReAct · AlfWorld | tasks solved: 130 out of 134 | Helped |
| add Reflexion | ReAct + gpt-4 · HotPotQA | accuracy: 0.39 -> 0.51 | Helped |
| add Reflexion | CoT (GT) + gpt-4 · HotPotQA | accuracy: 0.68 -> 0.80 | Helped |
| add self-generated test generation (ablation) | Reflexion · HumanEval Rust hardest 50 | pass@1: 0.52 -> 0.68 | Helped |
| add self-reflection step (ablation) | Reflexion · HumanEval Rust hardest 50 | pass@1: 0.60 -> 0.68 | Helped |
| add self-reflection on top of episodic memory of last trajectory (ablation) | CoT (GT) · HotPotQA | accuracy: 8% absolute | Helped |
| add Reflexion with weaker model | starchat-beta · HumanEval Python | pass@1: 0.26 -> 0.26 | No clear effect |
Closest papers by meaning
- Pushing Forward Pareto Frontiers of Proactive Agents with Behavioral Agentic Optimization (ICML 2026 · similarity 0.88)
- Ground-Compose-Reinforce: Grounding Language in Agentic Behaviours using Limited Data (NeurIPS 2025 · similarity 0.88)
- Representation-Based Exploration for Language Models: From Test-Time to Post-Training (ICLR 2026 · similarity 0.88)
- Controlling Large Language Model with Latent Action (ICML 2025 · similarity 0.88)
- Retroformer: Retrospective Large Language Agents with Policy Gradient Optimization (ICLR 2024 · similarity 0.88)
- Recursive Introspection: Teaching Language Model Agents How to Self-Improve (NeurIPS 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.