Harmonizing Dense and Sparse Signals in Multi-turn RL: Dual-Horizon Credit Assignment for Industrial Sales Agents (2026.acl-industry)
Copied to clipboard
| Challenge: | Large language models for industrial sales require balancing long-term commercial objectives with immediate linguistic constraints such as fluency and compliance. |
| Approach: | They propose a framework that disentangles optimization across time scales by normalizing advantages from turn-level and session-level rewards before fusion. |
| Outcome: | The proposed framework outperforms the state-of-the-art GRPO model in conversion rate and identity detection rate. |
Similar Papers
MASPO: Unifying Gradient Utilization, Probability Mass, and Signal Reliability for Robust and Sample-Efficient LLM Reasoning (2026.acl-long)
Copied to clipboard
Xiaoliang Fu, Jiaye Lin, Yangyi Fang, Binbin Zheng, Chaowen Hu, Zekai Shao, Cong Qin, Lu Pan, Ke Zeng, Xunliang Cai
| Challenge: | Existing RLVR algorithms rely on rigid, uniform, and symmetric trust region mechanisms . current algorithms lack robustness, asymmetric signal reliability and inefficient gradient utilization . |
| Approach: | They propose a framework to harmonize three dimensions of RLVR algorithms, a paper argues . a binary cutoff is used to discard valuable reinforcement signals, they argue . |
| Outcome: | The proposed framework outperforms baselines in evaluating a robust RLVR solution. |
Empowering Multi-Turn Tool-Integrated Agentic Reasoning with Group Turn Policy Optimization (2026.acl-long)
Copied to clipboard
Yifeng Ding, Hung Le, Songyang Han, Kangrui Ruan, Zhenghui Jin, Varun Kumar, Zijian Wang, Anoop Deoras
| Challenge: | Current reinforcement learning methods suffer from coarse-grained, trajectory-level rewards that provide insufficient learning signals for complex multi-turn interactions, leading to training stagnation. |
| Approach: | They propose a novel RL algorithm for training large language models for multi-turn tool-integrated reasoning (TIR) that incorporates three innovations: turn-level reward assignment that provides fine-grained feedback for individual turns, return-based advantage estimation where normalized discounted returns are calculated as advantages, and self-supervised reward shaping that exploits self-supervision signals from generated code to densify sparse binary outcome-based rewards. |
| Outcome: | The proposed algorithm outperforms GRPO by 3.0% across diverse math reasoning benchmarks and improves grepo by 3.9% on commonsense reasoning and program synthesis tasks. |
TSPO: Breaking the Double Homogenization Dilemma in Multi-turn Search Policy Optimization (2026.findings-acl)
Copied to clipboard
Shichao Ma, Zhiyuan Ma, Ming Yang, Xiaofan Li, Xing Wu, Jintao Du, Yu Cheng, Weiqiang Wang, Qiliang Liu, Zhengyang Zhou, Yang Wang
| Challenge: | Large Language Models (LLMs) can solve complex tasks through iterative information retrieval. |
| Approach: | They propose a turn-level stage-aware policy optimization approach to solve this problem . they introduce a first-occurrence latent reward mechanism to allocate partial rewards . |
| Outcome: | Experiments show that TSPO outperforms state-of-the-art models on Qwen2.5-3B and 7B models. |
Beyond the Context Window: Scaling Agentic RL via End-to-end Optimized Context Compression (2026.acl-long)
Copied to clipboard
| Challenge: | Existing reinforcement learning pipelines suffer from degraded instruction following, excessive rollout costs, and strict context limits. |
| Approach: | They propose a reinforcement learning (RL) fine-tuning of large language model (LLM) agents for long-horizon multi-turn tool use where context length quickly becomes a bottleneck. |
| Outcome: | The proposed framework improves the success rate while maintaining the same or even lower working context length compared to baselines. |
AT²PO: Agentic Turn-based Policy Optimization via Tree Search (2026.acl-long)
Copied to clipboard
| Challenge: | Recent advances in large language models (LLMs) have catalyzed the development of autonomous agents capable of executing complex, multi-turn tasks. |
| Approach: | They propose a framework for agentic reinforcement learning that integrates turn-level tree search with tree search to address key challenges. |
| Outcome: | The proposed framework addresses key challenges: limited exploration diversity, sparse credit assignment, and misaligned policy optimization. |
MGPO: Thinking with Images via Multi-Turn Grounding-Based Reinforcement Learning (2026.findings-acl)
Copied to clipboard
| Challenge: | State-of-the-art large multimodal models face challenges when processing high-resolution images, as these inputs are converted into enormous visual tokens, many of which are irrelevant to the downstream task. |
| Approach: | They propose a multi-turn grounding-based policy optimization framework that enables LMMs to iteratively focus on key visual regions by automatically cropping sub-images based on model-predicted grounding coordinates within a multiple-turn conversation framework. |
| Outcome: | The proposed framework improves on Qwen2.5-VL-7B with 21K samples and surpasses OpenAI’s o1 and GPT-4o models on the out-of-distribution (OOD) V* Bench. |
SALT: Step-level Advantage Assignment for Long-horizon Agents via Trajectory Graph (2026.findings-eacl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) have demonstrated remarkable capabilities, but their application to complex, multi-step, and long-horizon tasks remains challenging. |
| Approach: | They propose a framework that provides a finer-grained advantage assignment derived solely from outcome rewards. |
| Outcome: | The proposed framework provides a finer-grained advantage assignment, derived solely from outcome rewards. |
MatchTIR: Fine-Grained Supervision for Tool-Integrated Reasoning via Bipartite Matching (2026.acl-long)
Copied to clipboard
| Challenge: | Existing reinforcement learning methods rely on outcome- or trajectory-level rewards, assigning uniform advantages to all steps within a trajectory. |
| Approach: | They propose a framework that introduces fine-grained supervision via bipartite matching-based turn-level reward assignment and dual-level advantage estimation. |
| Outcome: | The proposed framework surpasses the majority of 8B competitors on three benchmarks. |
Beyond One-Preference-Fits-All Alignment: Multi-Objective Direct Preference Optimization (2024.findings-acl)
Copied to clipboard
| Challenge: | Recent approaches to language model alignment assume homogeneous human preferences, but actual human preferences vary widely and are hard to satisfy with a single language model. |
| Approach: | They propose an RL-free extension of Direct Preference Optimization (DPO) that folds language modeling directly into reward modeling and trains language models as collective reward models that combine all objectives with specific weights. |
| Outcome: | The proposed method matches or outperforms existing methods in safety alignment and long-form question answering. |
How to Allocate, How to Learn? Dynamic Rollout Allocation and Advantage Modulation for Policy Optimization (2026.findings-acl)
Copied to clipboard
Yangyi Fang, Jiaye Lin, Xiaoliang Fu, Cong Qin, Haolin Shi, Chaowen Hu, Lu Pan, Ke Zeng, Xunliang Cai
| Challenge: | Existing methods for reinforcement learning with verifiable rewards are limited by the complexity of the problem and the complexity. |
| Approach: | They propose a theoretically-grounded dual-pronged optimization framework for reinforcement learning with verifiable rewards that compensates for gradient attenuation of high-confidence correct actions while utilizing entropy changes as computable indicators to stabilize excessive update magnitudes. |
| Outcome: | The proposed framework compensates for gradient attenuation of high-confidence correct actions while utilizing entropy changes as computable indicators to stabilize excessive update magnitudes. |