# LOTAPO: Leave-One-Turn Attribution for Self-Generated Process Rewards in Multi-Turn Search Reasoning

arXiv: https://arxiv.org/abs/2607.13501 · HTML: https://mlsanity.com/paper/2607.13501/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Placed in Knowledge- and tool-augmented reasoning, one of 17 directions in LLM reasoning.
- Builds on 13 of its 34 references that are in the map.
- First posted on arXiv 15 Jul 2026.

## Placed in

- LLM reasoning / [Knowledge- and tool-augmented reasoning](https://mlsanity.com/lr/knowledge-and-tool-augmented-reasoning/)
- LLM alignment / RLHF / preference optimisation / [Process reward models for reasoning](https://mlsanity.com/alignment/process-reward-models-for-reasoning/)
- Synthetic data and self-improvement for LLMs / [Self-evolving search agents](https://mlsanity.com/synth/self-evolving-search-agents/)

## Builds on (most cited first, 13 of 13)

- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024)
- [Let's Verify Step by Step](https://mlsanity.com/paper/2305.20050/) (2023)
- [Text Embeddings by Weakly-Supervised Contrastive Pre-training](https://mlsanity.com/paper/2212.03533/) (2022)
- [Solving math word problems with process- and outcome-based feedback](https://mlsanity.com/paper/2211.14275/) (2022)
- [R1-Searcher: Incentivizing the Search Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2503.05592/) (2025)
- [ReSearch: Learning to Reason with Search for LLMs via Reinforcement Learning](https://mlsanity.com/paper/2503.19470/) (2025)
- [Reinforcing Multi-Turn Reasoning in LLM Agents via Fine-Grained Reward Structure and Credit Assignment](https://mlsanity.com/paper/2505.11821/) (2025)
- [OPID: On-Policy Skill Distillation for Agentic Reinforcement Learning](https://mlsanity.com/paper/2606.26790/) (2026)
- [Rubric-based On-policy Distillation](https://mlsanity.com/paper/2605.07396/) (2026)
- [IG-Search: Step-Level Information Gain Rewards for Search-Augmented Reasoning](https://mlsanity.com/paper/2604.15148/) (2026)
- [Optimizing Agentic Reasoning with Retrieval via Synthetic Semantic Information Gain Reward](https://mlsanity.com/paper/2602.00845/) (2026)
- [RewardFlow: Topology-Aware Reward Propagation on State Graphs for Agentic RL with Large Language Models](https://mlsanity.com/paper/2603.18859/) (2026)

## Built on by (most cited first, 5 of 5)

- [AttriMem: Attribution-Guided Process Feedback for Agent Memory Construction](https://mlsanity.com/paper/2607.21106/) (2026)
- [Critical-State RL: Diagnosing Trainable States for Multi-Turn Tool Use](https://mlsanity.com/paper/2609.24985/) (2026)
- [IGSD: Environment-Verified Hindsight Self-Distillation for Search Agents](https://mlsanity.com/paper/2609.32694/) (2026)
- [Dr.Credit: Rubric-Grounded Process Credit Assignment for Deep Research Agents](https://mlsanity.com/paper/2609.34296/) (2026)
- [C3: Rethinking Process Supervision via Confidence-Guided Credit Correction in Multi-Turn Reinforcement Learning](https://mlsanity.com/paper/2610.19695/) (2026)

## Abstract (arXiv)

Reinforcement learning for multi-turn search reasoning typically relies on terminal outcome rewards, which cannot distinguish useful, redundant, and harmful intermediate interactions. We propose LOTAPO , a self-generated process-supervision method based on backward leave-one-turn attribution. For each search turn, LOTAPO replaces the turn and its retrieval observation with a fixed [DELETE] placeholder and measures the resulting change in the current policy's mean log-likelihood of the gold answer. This Answer-Likelihood Gain estimates the turn's contribution while preserving all downstream interactions, allowing early evidence to be evaluated in the complete reasoning context. LOTAPO further applies sign-consistency gating, retaining only normalized process advantages whose directions agree with their raw attribution scores. The method requires no additional reward model, teacher, verifier, or LLM-as-a-Judge. Across seven knowledge-intensive question-answering datasets with local retrieval, LOTAPO achieves an average exact-match score of 0.326, outperforming the strongest step-reward baseline, IGPO, by 0.053. Ablations show complementary benefits from backward attribution and sign-consistency gating, demonstrating that policy-derived retrospective attribution can provide effective process supervision for multi-turn search agents.