Credit Where It Matters: Dependency-Aware Policy Optimization for Terminal Agents
cs.AI
Submitted: 2026-10-02
Updated: 2026-10-02
Code: https://github.com/google-gemini/gemini-cli
Sources
- AgentTrace: A Structured Logging Framework for Agent System Observability
- Beyond Next-Observation Prediction: Agent-Authored World Modeling for Sequential Decision Making
- Beyond Trajectory-Level Attribution: Graph-Based Credit Assignment for Agentic Reinforcement Learning
- Process Reinforcement through Implicit Rewards
- SWE-Shepherd: Advancing PRMs for Reinforcing Code Agents
- Endless Terminals: Scaling RL Environments for Terminal Agents
- SWE-TRACE: Optimizing Long-Horizon SWE Agents Through Rubric Process Reward Models and Heuristic Test-Time Scaling
- Tmax: A simple recipe for terminal agents
- CodeTracer: Towards Traceable Agent States
- PhGPO: Pheromone-Guided Policy Optimization for Long-Horizon Tool Planning
- Granularity-Adaptive Credit Assignment for Long-Horizon LLM Agent Reinforcement Learning
- Rethinking the Trust Region in LLM Reinforcement Learning
- Proximal Policy Optimization Algorithms
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- SETA: Scaling Environments for Terminal Agents
- Hindsight Credit Assignment for Long-Horizon LLM Agents
- From Agent Traces to Trust: A Survey of Evidence Tracing and Execution Provenance in LLM Agents
- Group-Graph Policy Optimization for Long-Horizon Agentic Reinforcement Learning
- T1: Terminal Agent Reinforcement Learning for Long-Horizon Tasks
- Learning from All-Failed Groups via First-Visit Coverage in Long-Horizon Agentic Tasks
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