본문으로 건너뛰기

#Reinforcement Learning

633개의 포스트

[논문리뷰] Watch Before You Answer: Learning from Visually Grounded Post-Training

본 논문은 오직 Visually Grounded 질문만을 사용하여 모델을 post-training하는 VidGround 프레임워크를 제안한다. 저자들은 GPT-5-mini와 같은 강력한 언어 모델을 사용하여 텍스트만으로 답변 가능한 TA 질문을 필터링하고, 나머지 VG 질문만을 학습 데이터로 선별하였다.
#Review#Vision-Language Models#Video Understanding#Post-Training#Linguistic Bias#Reinforcement Learning#Data Curation#Visually Grounded Reasoning
댓글 수 로딩 중

[논문리뷰] ThinkTwice: Jointly Optimizing Large Language Models for Reasoning and Self-Refinement

본 논문은 Reasoning 최적화와 Self-Refinement 최적화를 하나의 GRPO 프레임워크 안에서 결합한 ThinkTwice를 제안합니다. ThinkTwice는 각 훈련 단계에서 모델이 먼저 Reasoning 문제를 풀고, 동일한 문제에 대해 자신의 이전 답변을 개선하는(Thinking twice) 과정을 연속적으로 수행합니다 .
#Review#Large Language Models#Reinforcement Learning#Reasoning#Self-Refinement#RLVR#Policy Optimization#Implicit Curriculum
댓글 수 로딩 중

[논문리뷰] DARE: Diffusion Large Language Models Alignment and Reinforcement Executor

본 논문은 dLLM을 위한 통합 후학습 및 평가 프레임워크인 DARE (dLLMs Alignment and Reinforcement Executor)를 제안한다. DARE는 verl과 OpenCompass를 기반으로 하며, 다양한 모델 패밀리(MDLM 및 BDLM)를 동일한 실행 스택에서 처리할 수 있도록 설계되었다 .
#Review#Diffusion Large Language Models#Post-Training#Reinforcement Learning#Unified Framework#Systems Optimization
댓글 수 로딩 중

[논문리뷰] Vero: An Open RL Recipe for General Visual Reasoning

저자들은 6개 범주를 아우르는 600K 샘플의 Vero-600K를 구축하고, 태스크별로 세분화된 보상 함수를 적용하는 GSPO 기반의 단일 단계 RL 학습을 수행한다 . 데이터 정제 과정에서 모델 기반의 문항 필터링과 정답 정규화를 통해 학습 품질을 극대화하였다.
#Review#Vision-Language Models#Reinforcement Learning#Visual Reasoning#Multi-task Learning#Chain-of-Thought#Data Diversity#Reward Engineering
댓글 수 로딩 중

[논문리뷰] MemRerank: Preference Memory for Personalized Product Reranking

저자들은 사용자 구매 이력을 within-category와 cross-category 선호도로 구분하여 추출하는 Memory Extractor 모델을 설계하였다. 이 추출기는 재순위화 작업의 성능을 극대화하기 위해 GRPO 기반의 강화학습으로 post-training 되며, 이를 통해 별도의 Gold Label 없이도 재순위화 보상을 직접 최적화한다 .
#Review#Recommender Systems#Large Language Models#Product Reranking#Preference Memory#Reinforcement Learning
댓글 수 로딩 중

[논문리뷰] BeamPERL: Parameter-Efficient RL with Verifiable Rewards Specializes Compact LLMs for Structured Beam Mechanics Reasoning

arXiv에 게시된 'BeamPERL: Parameter-Efficient RL with Verifiable Rewards Specializes Compact LLMs for Structured Beam Mechanics Reasoning' 논문에 대한 자세한 리뷰입니다.
#Review#Reinforcement Learning#Parameter-Efficient Fine-Tuning (PEFT)#Large Language Models (LLM)#Beam Mechanics#Verifiable Rewards#Engineering Reasoning#Structural Engineering#Group Relative Policy Optimization (GRPO)
댓글 수 로딩 중

[논문리뷰] EgoPush: Learning End-to-End Egocentric Multi-Object Rearrangement for Mobile Robots

Sihang Li이 arXiv에 게시한 'EgoPush: Learning End-to-End Egocentric Multi-Object Rearrangement for Mobile Robots' 논문에 대한 자세한 리뷰입니다.
#Review#Egocentric Perception#Multi-Object Rearrangement#Mobile Robotics#Reinforcement Learning#Teacher-Student Distillation#Non-Prehensile Manipulation#Sim-to-Real Transfer#Object-Centric Representation
댓글 수 로딩 중

[논문리뷰] Back to Basics: Revisiting Exploration in Reinforcement Learning for LLM Reasoning via Generative Probabilities

Ivan Oseledets이 arXiv에 게시한 'Back to Basics: Revisiting Exploration in Reinforcement Learning for LLM Reasoning via Generative Probabilities' 논문에 대한 자세한 리뷰입니다.
#Review#Reinforcement Learning#LLM Reasoning#Exploration-Exploitation#Group Relative Policy Optimization#Entropy Collapse#Generative Models#Confidence-Aware Rewards
댓글 수 로딩 중

[논문리뷰] BatCoder: Self-Supervised Bidirectional Code-Documentation Learning via Back-Translation

Xiaohua Wang이 arXiv에 게시한 'BatCoder: Self-Supervised Bidirectional Code-Documentation Learning via Back-Translation' 논문에 대한 자세한 리뷰입니다.
#Review#Self-Supervised Learning#Code Generation#Documentation Generation#Back-Translation#Reinforcement Learning#Large Language Models (LLMs)#Code-Documentation Alignment#Low-Resource Languages
댓글 수 로딩 중

[논문리뷰] SkinFlow: Efficient Information Transmission for Open Dermatological Diagnosis via Dynamic Visual Encoding and Staged RL

arXiv에 게시된 'SkinFlow: Efficient Information Transmission for Open Dermatological Diagnosis via Dynamic Visual Encoding and Staged RL' 논문에 대한 자세한 리뷰입니다.
#Review#Dermatological Diagnosis#Multimodal LLM#Reinforcement Learning#Dynamic Visual Encoding#Information Transmission#Clinically Grounded Evaluation
댓글 수 로딩 중

[논문리뷰] CogFlow: Bridging Perception and Reasoning through Knowledge Internalization for Visual Mathematical Problem Solving

Tao Feng이 arXiv에 게시한 'CogFlow: Bridging Perception and Reasoning through Knowledge Internalization for Visual Mathematical Problem Solving' 논문에 대한 자세한 리뷰입니다.
#Review#Multimodal LLMs#Visual Reasoning#Mathematical Problem Solving#Knowledge Internalization#Reinforcement Learning#Cognitive-Inspired AI#Perception-Reasoning Alignment
댓글 수 로딩 중

[논문리뷰] Talk2Move: Reinforcement Learning for Text-Instructed Object-Level Geometric Transformation in Scenes

Shuo Yang이 arXiv에 게시한 'Talk2Move: Reinforcement Learning for Text-Instructed Object-Level Geometric Transformation in Scenes' 논문에 대한 자세한 리뷰입니다.
#Review#Reinforcement Learning#Text-Guided Image Editing#Object-Level Transformation#Geometric Transformation#Diffusion Models#GRPO#Scene Editing#Spatially Grounded Rewards
댓글 수 로딩 중

[논문리뷰] Let It Flow: Agentic Crafting on Rock and Roll, Building the ROME Model within an Open Agentic Learning Ecosystem

Wei Gao이 arXiv에 게시한 'Let It Flow: Agentic Crafting on Rock and Roll, Building the ROME Model within an Open Agentic Learning Ecosystem' 논문에 대한 자세한 리뷰입니다.
#Review#Agentic Learning Ecosystem#Large Language Models#Reinforcement Learning#Agentic Crafting#Tool Use#ROME Model#Policy Optimization#Sandbox Environment
댓글 수 로딩 중

[논문리뷰] ShowTable: Unlocking Creative Table Visualization with Collaborative Reflection and Refinement

Zhaohe Liao이 arXiv에 게시한 'ShowTable: Unlocking Creative Table Visualization with Collaborative Reflection and Refinement' 논문에 대한 자세한 리뷰입니다.
#Review#Table Visualization#Infographic Generation#Multi-modal Large Language Models (MLLMs)#Diffusion Models#Self-Correction#Reinforcement Learning#Graphic Design#Data-to-Visual Mapping
댓글 수 로딩 중

[논문리뷰] Decouple to Generalize: Context-First Self-Evolving Learning for Data-Scarce Vision-Language Reasoning

arXiv에 게시된 'Decouple to Generalize: Context-First Self-Evolving Learning for Data-Scarce Vision-Language Reasoning' 논문에 대한 자세한 리뷰입니다.
#Review#Vision-Language Models#Reinforcement Learning#Self-Evolving Learning#Data-Scarce Domains#Context-First Learning#Reward Hacking Mitigation#Multimodal Reasoning#Curriculum Learning
댓글 수 로딩 중

[논문리뷰] Beyond Token-level Supervision: Unlocking the Potential of Decoding-based Regression via Reinforcement Learning

Jiacheng Chen이 arXiv에 게시한 'Beyond Token-level Supervision: Unlocking the Potential of Decoding-based Regression via Reinforcement Learning' 논문에 대한 자세한 리뷰입니다.
#Review#Decoding-based Regression#Reinforcement Learning#Numerical Prediction#Large Language Models#Policy Gradient#Tokenization#Sequence Generation
댓글 수 로딩 중

[논문리뷰] From Imitation to Discrimination: Toward A Generalized Curriculum Advantage Mechanism Enhancing Cross-Domain Reasoning Tasks

Yang Li이 arXiv에 게시한 'From Imitation to Discrimination: Toward A Generalized Curriculum Advantage Mechanism Enhancing Cross-Domain Reasoning Tasks' 논문에 대한 자세한 리뷰입니다.
#Review#Reinforcement Learning#Large Language Models#Curriculum Learning#Advantage Function#Reasoning Tasks#Multimodal AI#Policy Optimization#Generalization
댓글 수 로딩 중

[논문리뷰] REVISOR: Beyond Textual Reflection, Towards Multimodal Introspective Reasoning in Long-Form Video Understanding

Jingyang Chen이 arXiv에 게시한 'REVISOR: Beyond Textual Reflection, Towards Multimodal Introspective Reasoning in Long-Form Video Understanding' 논문에 대한 자세한 리뷰입니다.
#Review#Multimodal Reasoning#Long-Form Video Understanding#Self-Reflection#Reinforcement Learning#Tool-Augmented MLLMs#Visual Rethinking#Video Question Answering#Causal Attribution
댓글 수 로딩 중

[논문리뷰] MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scaling

cyyang822이 arXiv에 게시한 'MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scaling' 논문에 대한 자세한 리뷰입니다.
#Review#Research Agent#Tool-Augmented Reasoning#Interaction Scaling#Large Language Models#Reinforcement Learning#Context Management#Open-Source AI
댓글 수 로딩 중

[논문리뷰] TimeSearch-R: Adaptive Temporal Search for Long-Form Video Understanding via Self-Verification Reinforcement Learning

arXiv에 게시된 'TimeSearch-R: Adaptive Temporal Search for Long-Form Video Understanding via Self-Verification Reinforcement Learning' 논문에 대한 자세한 리뷰입니다.
#Review#Long-form Video Understanding#Temporal Search#Reinforcement Learning#Self-Verification#Video-Language Models#Adaptive Search#Interleaved Reasoning
댓글 수 로딩 중

[논문리뷰] ChartM^3: A Multi-Stage Code-Driven Pipeline for Constructing Multi-Dimensional and Multi-Step Visual Reasoning Data in Chart Comprehension

Hao Wang이 arXiv에 게시한 'ChartM^3: A Multi-Stage Code-Driven Pipeline for Constructing Multi-Dimensional and Multi-Step Visual Reasoning Data in Chart Comprehension' 논문에 대한 자세한 리뷰입니다.
#Review#Chart Comprehension#Visual Reasoning#Data Generation#Code-Driven Pipeline#Multimodal LLMs#Retrieval-Augmented Generation#Reinforcement Learning#Synthetic Data
댓글 수 로딩 중

[논문리뷰] Unleashing Scientific Reasoning for Bio-experimental Protocol Generation via Structured Component-based Reward Mechanism

Shuang Gu이 arXiv에 게시한 'Unleashing Scientific Reasoning for Bio-experimental Protocol Generation via Structured Component-based Reward Mechanism' 논문에 대한 자세한 리뷰입니다.
#Review#Scientific Reasoning#Bio-experimental Protocol Generation#LLM#Structured Reward#SciRecipe Dataset#Sketch-and-Fill#Reinforcement Learning#Thoth
댓글 수 로딩 중

[논문리뷰] A Goal Without a Plan Is Just a Wish: Efficient and Effective Global Planner Training for Long-Horizon Agent Tasks

Fanchao Qi이 arXiv에 게시한 'A Goal Without a Plan Is Just a Wish: Efficient and Effective Global Planner Training for Long-Horizon Agent Tasks' 논문에 대한 자세한 리뷰입니다.
#Review#Long-Horizon Tasks#LLM Agents#Global Planning#Reinforcement Learning#Supervised Fine-tuning#Homologous Consensus Filtering#Executor Capability Gain Reward#Plan-and-Execute
댓글 수 로딩 중

[논문리뷰] MM-HELIX: Boosting Multimodal Long-Chain Reflective Reasoning with Holistic Platform and Adaptive Hybrid Policy Optimization

vanilla1116이 arXiv에 게시한 'MM-HELIX: Boosting Multimodal Long-Chain Reflective Reasoning with Holistic Platform and Adaptive Hybrid Policy Optimization' 논문에 대한 자세한 리뷰입니다.
#Review#Multimodal LLMs#Reflective Reasoning#Long-Chain Reasoning#Benchmark#Policy Optimization#Data Generation#Reinforcement Learning#Backtracking
댓글 수 로딩 중

[논문리뷰] Entropy Regularizing Activation: Boosting Continuous Control, Large Language Models, and Image Classification with Activation as Entropy Constraints

Huazhe Xu이 arXiv에 게시한 'Entropy Regularizing Activation: Boosting Continuous Control, Large Language Models, and Image Classification with Activation as Entropy Constraints' 논문에 대한 자세한 리뷰입니다.
#Review#Entropy Regularization#Activation Functions#Continuous Control#Large Language Models#Image Classification#Reinforcement Learning#Policy Stochasticity#Entropy Constraints
댓글 수 로딩 중

[논문리뷰] Learn the Ropes, Then Trust the Wins: Self-imitation with Progressive Exploration for Agentic Reinforcement Learning

Gang Li이 arXiv에 게시한 'Learn the Ropes, Then Trust the Wins: Self-imitation with Progressive Exploration for Agentic Reinforcement Learning' 논문에 대한 자세한 리뷰입니다.
#Review#Reinforcement Learning#LLM Agents#Exploration-Exploitation#Self-Imitation Learning#Intrinsic Rewards#Curriculum Learning#Policy Entropy#Tool Use
댓글 수 로딩 중

[논문리뷰] The Choice of Divergence: A Neglected Key to Mitigating Diversity Collapse in Reinforcement Learning with Verifiable Reward

Xiaoyu Tan이 arXiv에 게시한 'The Choice of Divergence: A Neglected Key to Mitigating Diversity Collapse in Reinforcement Learning with Verifiable Reward' 논문에 대한 자세한 리뷰입니다.
#Review#Reinforcement Learning#Large Language Models (LLMs)#Diversity Collapse#f-divergence#Forward-KL#JS-divergence#Pass@k#Catastrophic Forgetting
댓글 수 로딩 중

[논문리뷰] AgentGym-RL: Training LLM Agents for Long-Horizon Decision Making through Multi-Turn Reinforcement Learning

Honglin Guo이 arXiv에 게시한 'AgentGym-RL: Training LLM Agents for Long-Horizon Decision Making through Multi-Turn Reinforcement Learning' 논문에 대한 자세한 리뷰입니다.
#Review#LLM Agents#Reinforcement Learning#Multi-Turn Interaction#Long-Horizon Decision Making#Agent Framework#Exploration-Exploitation#Progressive Scaling
댓글 수 로딩 중

[논문리뷰] CODA: Coordinating the Cerebrum and Cerebellum for a Dual-Brain Computer Use Agent with Decoupled Reinforcement Learning

Jianze Liang이 arXiv에 게시한 'CODA: Coordinating the Cerebrum and Cerebellum for a Dual-Brain Computer Use Agent with Decoupled Reinforcement Learning' 논문에 대한 자세한 리뷰입니다.
#Review#GUI Agents#Reinforcement Learning#Planner-Executor Architecture#Decoupled Training#Large Vision-Language Models#Specialization#Generalization#Computer Use Agent
댓글 수 로딩 중

[논문리뷰] TreePO: Bridging the Gap of Policy Optimization and Efficacy and Inference Efficiency with Heuristic Tree-based Modeling

Zhoufutu Wen이 arXiv에 게시한 'TreePO: Bridging the Gap of Policy Optimization and Efficacy and Inference Efficiency with Heuristic Tree-based Modeling' 논문에 대한 자세한 리뷰입니다.
#Review#Reinforcement Learning#Policy Optimization#Large Language Models#Inference Efficiency#Tree Search#Segment-level Decoding#Advantage Estimation#Reasoning
댓글 수 로딩 중

[논문리뷰] Breaking the Exploration Bottleneck: Rubric-Scaffolded Reinforcement Learning for General LLM Reasoning

Jiale Zhao이 arXiv에 게시한 'Breaking the Exploration Bottleneck: Rubric-Scaffolded Reinforcement Learning for General LLM Reasoning' 논문에 대한 자세한 리뷰입니다.
#Review#Reinforcement Learning#Large Language Models#Exploration Bottleneck#Instructional Scaffolding#Rubric-based Rewards#General Reasoning#RL with Verifiable Rewards#Policy Optimization
댓글 수 로딩 중

[논문리뷰] On-Policy RL Meets Off-Policy Experts: Harmonizing Supervised Fine-Tuning and Reinforcement Learning via Dynamic Weighting

Guoyin Wang이 arXiv에 게시한 'On-Policy RL Meets Off-Policy Experts: Harmonizing Supervised Fine-Tuning and Reinforcement Learning via Dynamic Weighting' 논문에 대한 자세한 리뷰입니다.
#Review#Large Language Models#Reinforcement Learning#Supervised Fine-Tuning#On-Policy RL#Off-Policy Experts#Dynamic Weighting#LLM Alignment#Reasoning
댓글 수 로딩 중

[논문리뷰] RL-PLUS: Countering Capability Boundary Collapse of LLMs in Reinforcement Learning with Hybrid-policy Optimization

Kechi Zhang이 arXiv에 게시한 'RL-PLUS: Countering Capability Boundary Collapse of LLMs in Reinforcement Learning with Hybrid-policy Optimization' 논문에 대한 자세한 리뷰입니다.
#Review#Large Language Models#Reinforcement Learning#Capability Collapse#Hybrid Policy Optimization#Multiple Importance Sampling#Exploration#Math Reasoning#Out-of-Distribution
댓글 수 로딩 중