Challenge: Logic-RL is a framework that transforms critique-guided outline refinement into a learnable policy through reinforcement learning.
Approach: They propose a framework that transforms critique-guided outline refinement into a learnable policy through reinforcement learning.
Outcome: The proposed framework improves on FreshWiki and WikiOutline . it can be iteratively applied, with improved quality continuing through three refinement rounds before diminishing returns.

Similar Papers

Logic: Long-form Outline Generation via Imitative and Critical Self-refinement (2025.findings-emnlp)

Copied to clipboard

Challenge: Existing methods for long-form outline generation have low knowledge density and lack detail . retrieval-augmented approaches struggle to maintain logical coherence across retrieved information .
Approach: They propose a system that mimics human writers' refinement process by mimicking outlines through imitation and critical self-refinement.
Outcome: The proposed system improves on the FreshWiki and WikiOutline datasets and establishes a coherent planning framework and structured knowledge base.
Training Language Model to Critique for Better Refinement (2025.findings-acl)

Copied to clipboard

Challenge: Large language models (LLMs) have remarkable evaluation and critique capabilities, providing insightful feedback and identifying flaws in various tasks.
Approach: They propose a framework to train critic models using refinement signals to generate feedback loops where critiques guide the model in refining its responses.
Outcome: The proposed framework outperforms traditional methods and open-source models in terms of critique quality and refinement outcomes.
Learning to Refine with Fine-Grained Natural Language Feedback (2024.findings-emnlp)

Copied to clipboard

Challenge: Recent work has explored the capability of large language models to identify and correct errors in LLM-generated responses.
Approach: They propose to combine refinement with feedback into three distinct competencies . step 1: Detect, Critique, Refine gives a fine-grained feedback about errors .
Outcome: The proposed method outperforms existing refinement approaches and models not fine-tuned for factuality critiquing.
Learning to Refine: Self-Refinement of Parallel Reasoning in LLMs (2026.findings-acl)

Copied to clipboard

Challenge: Existing approaches to test-time scaling are limited due to the quality of candidate responses.
Approach: They propose a new metric to quantify the relative improvement of self-refinement beyond majority voting.
Outcome: The proposed method achieves state-of-the-art performance across five benchmarks over other methods.
TemplateRL: Structured Template-Guided Reinforcement Learning for LLM Reasoning (2026.findings-acl)

Copied to clipboard

Challenge: Existing RL methods rely on unstructured self-sampling to fit scalar rewards, resulting in inefficient rollouts.
Approach: They propose a structured template-guided RL framework that augments policy optimization with explicit template guidance.
Outcome: Experiments show that TemplateRL outperforms GRPO and GRPI by 99% on AIME and 41% on AMC with superior stability on weak models and remarkable cross-domain generalization.
Learning from Cognition: Enhancing RL Efficiency for LLM Reasoning via Hierarchical Metacognitive Decomposition and Refinement (2026.acl-long)

Copied to clipboard

Challenge: Recent advances in Large Language Models have demonstrated notable inferential capacities via reinforcement learning (RL) however, “zero-RL” approaches relying on fixed prompt templates introduce substantial sampling inefficiencies for weak LLMs.
Approach: They propose a hierarchical metacognitive RL framework that decomposes zero-accuracy problems into subproblems and prompts the policy to refine answers by referencing previous wrong solutions.
Outcome: The proposed framework improves sample utilization and sample efficiency and accelerates convergence compared to baselines.
Enhancing Reinforcement Learning with Dense Rewards from Language Model Critic (2024.emnlp-main)

Copied to clipboard

Challenge: Reinforcement learning (RL) can align language models with non-differentiable reward signals, such as human preferences, but the sparsity of these signals can lead to inefficient and unstable learning.
Approach: They propose a framework that utilizes the critique capability of Large Language Models to produce intermediate-step rewards during RL training.
Outcome: The proposed framework improves sample efficiency and the overall performance of the policy model, supported by both automatic and human evaluation.
Arg-LLaDA: Argument Summarization via Large Language Diffusion Models and Sufficiency-Aware Refinement (2026.acl-long)

Copied to clipboard

Challenge: Existing approaches to argument summarization rely on single-pass generation, offering limited support for factual correction or structural refinement.
Approach: They propose a large language diffusion framework that iteratively improves argument summarization by sufficiency-guided remasking and regeneration.
Outcome: Empirical results show that Arg-LLaDA surpasses state-of-the-art baselines in 7 out of 10 evaluation metrics.
ToolDNA: Autonomous Evolution of Tool Metadata for Robust Dialogue Agents (2026.findings-acl)

Copied to clipboard

Challenge: Task-oriented dialogue systems face labor-intensive manual metadata tuning and sparse reinforcement learning (RL) rewards that fail to diagnose invocation errors.
Approach: They propose a framework that enables auto-evolution of policy networks and tool metadata via RL . a tool metadata loop coordinates metadata through policy-generated candidates during rollouts .
Outcome: The proposed framework achieves +11% problem resolution and +54% accuracy over commercial LLMs with prompt engineering and +25%/+35% over supervised fine-tuning.
CURE: Critique-Driven Unified Reinforcement Learning for Test-Time Self-Improvement (2026.acl-long)

Copied to clipboard

Challenge: Existing critique-guided methods fail to equip models with the autonomous improvement capabilities required for test-time scaling.
Approach: They propose a framework that jointly optimizes a single policy for standard solving, critiquing, and guided re-exploration.
Outcome: The proposed framework maintains competitive single-turn performance and unlocks effective inference-time scaling.

What is GenGO?

GenGO is an NLP powered publication search system. It currenctly indexes 30k+ papers from ACL Anthology, and implements multi-aspect summarization, semantic search, and more!

Information

About
Limitations