[Agent] ToolRL: Reward is All Tool Learning Needs — tool-use RL에서 reward 설계만 바꿔도 되는가
[Agent] ToolRL: Reward is All Tool Learning Needs
Recent posts
[Agent] SynPlanResearch-R1: 랜덤 tool plan + cue로 합성한 cold-start 데이터가 RL의 탐색 천장을 올린다
[Agent] SynPlanResearch-R1: Encouraging Tool Exploration for Deep Research with Synthetic Plans
[Agent] RAGEN: Understanding Self-Evolution in LLM Agents via Multi-Turn RL — agent RL은 왜 무너지는가
[Agent] RAGEN: Understanding Self-Evolution in LLM Agents via Multi-Turn Reinforcement Learning
[Agent] IterCAD: 도면을 보고 고쳐 그리는 closed-loop CAD 코드 생성 에이전트
[Agent] IterCAD: An Iterative Multimodal Agent for Visually-Grounded CAD Generation and Editing
[VLM] Visual Preference Optimization with Rubric Rewards (rDPO): instance-specific rubric로 on-policy 선호쌍을 만드는 DPO 파이프라인
[VLM] Visual Preference Optimization with Rubric Rewards