Research map / Large language models / Agents, code and math
ReIn: Conversational Error Recovery with Reasoning Inception
Takyoung Kim, Jinseok Nam, Chandrayee Basu, Xing Fan, Chengyuan Ma, Heng Ji, Gokhan Tur, Dilek Hakkani-Tür
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 ReIn inception block | task agent (Sonnet 3.7, Haiku 3.5) · tau-Bench retail, seen error types | Pass^1 | Improved |
| add ReIn on ambiguous requests | task agent · tau-Bench ambiguous scenarios | Pass^1 | Improved |
| add ReIn on unseen error types (Contradiction, Domain) | task agent · tau-Bench unseen scenarios | Pass^1 | Improved |
| use smaller Llama 3.2 3B inception module instead of larger modules (ablation) | ReIn · tau-Bench seen and unseen | Pass^1: 100% -> 81.48% (activation, airline Anaphora) | Hurt |
| use ReIn instead of explicit prompt modification (NPI, self-refine) | Sonnet 3.7 · tau-Bench retail | Pass^1 | Improved |
| let ReIn activate dynamically at any turn | Sonnet 3.7 · tau-Bench airline | Pass^1 | Improved |
| define recovery plan as plain response phrase instead of a recovery tool (ablation) | ReIn · instruction-hierarchy test | task completion rate: tool-based -> 0% | Hurt |
Closest papers by meaning
- RefactorBench: Evaluating Stateful Reasoning in Language Agents Through Code (ICLR 2025 · similarity 0.88)
- Talk, Evaluate, Diagnose: User-aware Agent Evaluation with Automated Error Analysis (ICLR 2026 · similarity 0.87)
- Beyond Prompts: Dynamic Conversational Benchmarking of Large Language Models (NeurIPS 2024 · similarity 0.87)
- Memory-T1: Reinforcement Learning for Temporal Reasoning in Multi-session Agents (ICLR 2026 · similarity 0.86)
- $\tau^2$-Bench: Evaluating Conversational Agents in a Dual-Control Environment (ICML 2026 · similarity 0.86)
- Recon: Reconstruction-Guided Reasoning Synthesis for User Modeling (NeurIPS 2026 · similarity 0.86)
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.
