#Agentic Reasoning

12개의 포스트

[논문리뷰] LightThinker++: From Reasoning Compression to Memory Management

저자들은 암시적 압축에서 시작하여 명시적 행동 수준의 관리로 진화하는 LightThinker 계열 모델을 제안한다. LightThinker는 gist tokens와 특수 설계된 attention mask를 활용하여 긴 사고 과정을 컴팩트한 표현으로 변환한다 .

#Review #Large Language Models #Reasoning Compression #Memory Management #Agentic Reasoning #Context Optimization

2026년 4월 6일

[논문리뷰] Strategic Navigation or Stochastic Search? How Agents and Humans Reason Over Document Collections

arXiv에 게시된 'Strategic Navigation or Stochastic Search? How Agents and Humans Reason Over Document Collections' 논문에 대한 자세한 리뷰입니다.

#Review #Multimodal Agents #Document QA #Agentic Reasoning #RAG #Benchmark #PDFs #Effort Calibration

2026년 3월 12일

[논문리뷰] Experiential Reinforcement Learning

arXiv에 게시된 'Experiential Reinforcement Learning' 논문에 대한 자세한 리뷰입니다.

#Review #Reinforcement Learning #Language Models #Self-Reflection #Experiential Learning #Policy Optimization #Distillation #Agentic Reasoning

2026년 2월 16일

[논문리뷰] Agentic Reasoning for Large Language Models

arXiv에 게시된 'Agentic Reasoning for Large Language Models' 논문에 대한 자세한 리뷰입니다.

#Review #Agentic Reasoning #LLM Agents #Self-Evolving AI #Multi-Agent Systems #Planning #Tool Use #Retrieval-Augmented Generation #Reinforcement Learning

2026년 1월 21일

[논문리뷰] SenseNova-MARS: Empowering Multimodal Agentic Reasoning and Search via Reinforcement Learning

arXiv에 게시된 'SenseNova-MARS: Empowering Multimodal Agentic Reasoning and Search via Reinforcement Learning' 논문에 대한 자세한 리뷰입니다.

#Review #Multimodal Agents #Reinforcement Learning #Vision-Language Models #Tool Use #Agentic Reasoning #Image Search #HR-MMSearch #BN-GSPO

2026년 1월 4일

[논문리뷰] Nemotron 3 Nano: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning

arXiv에 게시된 'Nemotron 3 Nano: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning' 논문에 대한 자세한 리뷰입니다.

#Review #Mixture-of-Experts #Mamba-Transformer #Agentic Reasoning #Long Context LLM #FP8 Quantization #Supervised Fine-Tuning #Reinforcement Learning

2025년 12월 24일

[논문리뷰] Native Parallel Reasoner: Reasoning in Parallelism via Self-Distilled Reinforcement Learning

arXiv에 게시된 'Native Parallel Reasoner: Reasoning in Parallelism via Self-Distilled Reinforcement Learning' 논문에 대한 자세한 리뷰입니다.

#Review #Large Language Models (LLMs)#Parallel Reasoning #Self-Distilled Reinforcement Learning #Policy Optimization #Inference Acceleration #Structured Output #Agentic Reasoning

2025년 12월 8일

[논문리뷰] CodeV: Code with Images for Faithful Visual Reasoning via Tool-Aware Policy Optimization

arXiv에 게시된 'CodeV: Code with Images for Faithful Visual Reasoning via Tool-Aware Policy Optimization' 논문에 대한 자세한 리뷰입니다.

#Review #Vision-Language Models #Agentic Reasoning #Tool Use #Reinforcement Learning #Faithfulness Evaluation #Policy Optimization #Visual Search #Code Generation

2025년 12월 2일

[논문리뷰] TIR-Bench: A Comprehensive Benchmark for Agentic Thinking-with-Images Reasoning

Shaoheng Lin이 arXiv에 게시한 'TIR-Bench: A Comprehensive Benchmark for Agentic Thinking-with-Images Reasoning' 논문에 대한 자세한 리뷰입니다.

#Review #Multimodal LLMs #Agentic Reasoning #Thinking-with-Images #Visual Reasoning Benchmark #Tool Use #Image Manipulation #Fine-tuning

2025년 11월 9일

[논문리뷰] DeepAgent: A General Reasoning Agent with Scalable Toolsets

Jiajie Jin이 arXiv에 게시한 'DeepAgent: A General Reasoning Agent with Scalable Toolsets' 논문에 대한 자세한 리뷰입니다.

#Review #Autonomous Agents #Large Language Models #Tool Use #Reinforcement Learning #Memory Management #Tool Retrieval #Agentic Reasoning

2025년 10월 27일

[논문리뷰] AlphaApollo: Orchestrating Foundation Models and Professional Tools into a Self-Evolving System for Deep Agentic Reasoning

Zongze Li이 arXiv에 게시한 'AlphaApollo: Orchestrating Foundation Models and Professional Tools into a Self-Evolving System for Deep Agentic Reasoning' 논문에 대한 자세한 리뷰입니다.

#Review #Foundation Models #Agentic Reasoning #Tool Use #Self-Evolving System #Retrieval-Augmented Generation #Computational Tools #Error Correction

2025년 10월 9일

[논문리뷰] PVPO: Pre-Estimated Value-Based Policy Optimization for Agentic Reasoning

Yuewei Zhang이 arXiv에 게시한 'PVPO: Pre-Estimated Value-Based Policy Optimization for Agentic Reasoning' 논문에 대한 자세한 리뷰입니다.

#Review #Reinforcement Learning #Critic-Free RL #Agentic Reasoning #Policy Optimization #Large Language Models (LLMs)#Advantage Estimation #Group Sampling #Static Value Estimation

2025년 9월 2일