arXiv AI
Post-Training Large Language Models via Reinforcement Learning from Self-Feedback
Read the original on arXiv AI →The Flow has not summarised this story yet — read it at arXiv AI.
Related stories
arXiv AI
Sep 30
Unlocking the Critic: Reward-Free Policy Optimization for LLM Post-Training
arXiv:2609.37119v1 Announce Type: cross Abstract: Recent approaches to reinforcement learning (RL) post-training for large language models increasingly remove the critic to reduce training instabilit...
arXiv Machine Learning
Aug 11
Process Supervision of Confidence Margin for Calibrated LLM Reasoning
arXiv:2604. 23333v2 Announce Type: replace Abstract: Scaling test-time computation with reinforcement learning (RL) has emerged as a reliable path to improve large language models (LLM) reasoning ability.
arXiv Computation and Language
Oct 26
Free Lunch for LLMs Confidence: The RL Critic Already Knows
arXiv:2610.31820v1 Announce Type: new Abstract: Reliable confidence estimation is essential for trustworthy deployment of large language models (LLMs), but dedicated confidence training stage require...
arXiv Machine Learning
Sep 1
ERR+: Sequential Entropy Resolution for Efficient and Decisive LLM Reasoning
arXiv:2608.28771v1 Announce Type: new Abstract: Large reasoning models achieve strong performance on complex tasks by generating extended chain-of-thought (CoT) traces via reinforcement learning with...
arXiv AI
Jun 11
ProcessThinker: Enhancing Multi-modal Large Language Models Reasoning via Rollout-based Process Reward
arXiv:2606. 11209v1 Announce Type: cross Abstract: Visual question answering increasingly requires multi-step reasoning.
arXiv AI
Oct 5
Test-time Calibration Learning for Large Language Model Reasoning
The paper introduces Test-Time Calibration Learning (TTCL), a label‑free framework that adapts both reasoning accuracy and verbalized confidence of large language models directly on unlabeled target‑task data. TTCL generates self‑supervision signals from multiple model responses, enabling calibration without ground‑truth labels and proving theoretically as a bounded surrogate for the ideal calibration objective. Experiments on mathematical reasoning and factual question answering show consistent improvements, with base models gaining an average 40.13% in accuracy and 70.80% in ECE reduction across eight benchmarks, and further gains under domain shift.