SkillRubric: Co-Evolving Actor Guidance and Evaluator Rubrics for Multimodal Agents
cs.AI
Submitted: 2026-09-28
Updated: 2026-09-28
Terminology
Sources
- Qwen3-VL Technical Report
- CUA-Skill: Develop Skills for Computer Using Agent
- EvoRubrics: Dynamic Rubrics as Rewards via Adversarial Co-Evolution for LLM Reinforcement Learning
- Beyond Seeing: Evaluating Multimodal LLMs on Tool-Enabled Image Perception, Transformation, and Reasoning
- ReSkill: Reconciling Skill Creation with Policy Optimization in Agentic RL
- RubricEM: Meta-RL with Rubric-guided Policy Decomposition beyond Verifiable Rewards
- TIR-Bench: A Comprehensive Benchmark for Agentic Thinking-with-Images Reasoning
- ARES: Automated Rubric Synthesis for Scalable LLM Reinforcement Learning
- SkillGraph: Skill-Augmented Reinforcement Learning for Agents via Evolving Skill Graphs
- Self-Distilled Agentic Reinforcement Learning
- SKILL0: In-Context Agentic Reinforcement Learning for Skill Internalization
- Proximal Policy Optimization Algorithms
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- AgentVista: Evaluating Multimodal Agents in Ultra-Challenging Realistic Visual Scenarios
- TRACE: Turn-level Reward Assignment via Credit Estimation for Long-Horizon Agents
- MMSearch-Plus: Benchmarking Provenance-Aware Search for Multimodal Browsing Agents
- ARCO: Adaptive Rubrics with Co-Evolution for Multi-Step LLM-Based Agents
- Skill-SD: Skill-Conditioned Self-Distillation for Multi-turn LLM Agents
- RTMC: Step-Level Credit Assignment via Rollout Trees
- AgentOPSD: Recursive Self-Distillation for Agentic Reinforcement Learning
Related papers
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Agents' Last Exam
- The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribing
- MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Intervention
- Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systems