4arXiv cs.LG (Machine Learning)·
Semantic Pareto-DQN: Multi-Objective RL with LLM State Representations for Financial Fraud Detection
Researchers propose Semantic Pareto-DQN, a multi-objective reinforcement learning framework that addresses class imbalance in financial anomaly detection without data resampling. The approach encodes heterogeneous transaction features as natural-language narratives via LLMs to produce scale-invariant state representations, then optimizes a vectorial reward that decouples fraud detection, false-positive friction, and semantic discovery across the Pareto frontier. Empirical results on E-Commerce fraud and UCI Credit datasets show improved minority-class recall over scalarized baselines, avoiding the 'fraud collapse' failure mode.
Related guides (1)
Related events (8)
Two is better than one: A Collapse-free Multi-Reward RLIF Training Framework
This paper proposes a multi-reward reinforcement learning from internal feedback (RLIF) framework that decomposes training signals into an answer-level reward via cluster voting and a completion-level reward via token-wise self-certainty. To address reward hacking and entropy collapse common in single-reward RLIF, the authors introduce GDPO-based normalization and KL-Cov regularization targeting low-entropy token distributions. Evaluated on mathematical reasoning and code-generation benchmarks, the method achieves stability and performance approaching supervised RLVR methods without requiring external ground-truth supervision. The work advances scalable unsupervised RL training for LLM reasoning.
LamPO: Lambda-Style Policy Optimization with Pairwise Decomposed Advantage for Reasoning LMs
LamPO proposes a new RLVR training objective that replaces GRPO's scalar group-relative advantages with a Pairwise Decomposed Advantage, aggregating pairwise reward gaps within response groups and weighting comparisons by confidence-aware log-probability differences. The method retains a critic-free, clipped-update PPO-style structure and optionally adds a ROUGE-L-based dense auxiliary reward to reduce sparsity. Experiments on AIME24, AIME25, MATH-500, and GPQA-Diamond using Qwen3-1.7B, Qwen3-4B, and Phi-4-mini show consistent improvements over GRPO and other RLVR variants with more stable training dynamics.
Taiji: Pareto Optimal Policy Optimization for LLM-enhanced recommendation at Kuaishou scale
Researchers from Kuaishou present Taiji, an LLM-as-Enhancer framework for industrial recommender systems that addresses two bottlenecks: generating high-quality chain-of-thought data via reverse-engineered reasoning and rejection sampling during SFT, and balancing semantic vs. ID-based rewards during RL alignment via a new algorithm called Pareto Optimal Policy Optimization (POPO). The system has been deployed on Kuaishou's advertising platform since May 2026, serving over 400 million daily users. The paper contributes both a practical deployment case study and a novel RL alignment technique for the LLM4Rec paradigm.
Adaptive Data Scheduling (ADS) improves LLM reinforcement learning post-training by 5.2% over GRPO
Researchers propose Adaptive Data Scheduling (ADS), a dual-level framework that replaces uniform sampling in RL post-training with adaptive distribution over semantic clusters and policy-boundary sample selection. Evaluated across three LLMs and seven reasoning benchmarks, ADS improves average accuracy by 5.2% over GRPO and generalizes across RL objectives. The method addresses a structural limitation in standard RL post-training pipelines by accounting for semantic data structure and evolving policy capability during training.
MADA-RL: Multi-agent debate with counterfactual critic advantage for parameter-efficient reasoning in compact models
MADA-RL is a post-training framework that specializes compact LLMs (≤4B parameters) into generator and critic roles, training them with a debate-aware RL signal using LoRA adapters on only a small fraction of parameters. The central contribution is a counterfactual critic advantage that conditions the critic's reward on the generator ensemble's per-instance accuracy, explicitly incentivizing critics to correct generator errors rather than reproduce correct answers. Applied to DeepSeek-R1-Distill-Qwen-1.5B, the method achieves +2.0 percentage points on mathematical reasoning benchmarks using 16× fewer trainable parameters than fully fine-tuned baselines, placing it on the accuracy-vs-trainable-parameter Pareto front. The approach does not surpass the strongest baselines (DeepScaleR, STILL-3) trained on larger datasets, and the paper analyzes this gap directly.
Progress Advantage: Annotation-Free Step-Level Scoring for LLM Agents via RL Post-Training
Researchers introduce 'progress advantage,' a method that derives implicit step-level reward signals for LLM agents directly from the log-probability ratio between an RL-trained policy and its reference policy, without requiring dedicated process reward model training. The approach is shown to recover the optimal advantage function under a general stochastic MDP formulation, making it annotation-free and domain-agnostic. Validated across five benchmarks and four model families on tasks including test-time scaling, uncertainty quantification, and failure attribution, it outperforms confidence-based baselines and even dedicated trained reward models. The result is practically significant because building process reward models for agentic settings is currently a major bottleneck.
CARE-PPO: PPO-based RL framework for joint quantitative prediction and confidence estimation in LLMs
Researchers introduce CARE-PPO, a reinforcement learning fine-tuning framework that jointly trains LLMs for numerical prediction accuracy and calibrated confidence estimation. The approach repurposes the PPO critic as a confidence estimator at inference time, using a Confidence-Aligned Reward for Estimation derived from prediction error. Evaluated on healthcare and finance tasks with Qwen-3 4B and 8B models, CARE-PPO outperforms logit-based and verbalized confidence baselines and shows improved out-of-distribution generalization. The work addresses the hallucination and overconfidence problems that limit LLM deployment in high-stakes quantitative domains.
STARE: Token-level advantage reweighting to prevent entropy collapse in GRPO-style RL training
Researchers introduce STARE, a method addressing policy entropy collapse in GRPO-style reinforcement learning from verifiable rewards (RLVR) for LLM post-training. Through first-order gradient analysis, they identify a token-level credit assignment mismatch and propose selectively reweighting advantages for entropy-critical tokens using batch-internal surprisal quantiles plus a closed-loop entropy gate. Evaluated across 1.5B–32B models on short/long chain-of-thought and multi-turn tool use tasks, STARE outperforms DAPO and other baselines by 4–8% on AIME24/25 while sustaining stable training over thousands of steps.