Counterfactual Distillation: Internalizing Reflective Experience into LLM Agent
cs.AI
Submitted: 2026-03-17
Updated: 2026-11-06
Code: https://github.com/mpSchrader/gym-sokoban
Terminology
Sources
- Self-Improving LLM Agents at Test-Time
- Large Language Models for Mathematical Reasoning: Progresses and Challenges
- CodeIt: Self-Improving Language Models with Prioritized Hindsight Replay
- FLEX: Continuous Agent Evolution via Forward Learning from Experience
- Why Do LLM Agents Fail in Exploring New Environments? A World-Modeling Perspective
- Agent-R1: A Unified and Modular Framework for Agentic Reinforcement Learning
- Process Reinforcement through Implicit Rewards
- Agent-RLVR: Training Software Engineering Agents via Guidance and Environment Rewards
- Group-in-Group Policy Optimization for LLM Agent Training
- AReaL: A Large-Scale Asynchronous Reinforcement Learning System for Language Reasoning
- Deep Think with Confidence
- The Llama 3 Herd of Models
- DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
- World Models
- Mastering Atari with Discrete World Models
- Mastering Diverse Domains through World Models
- Skywork Open Reasoner 1 Technical Report
- EvoTest: Evolutionary Test-Time Learning for Self-Improving Agentic Systems
- Open-Reasoner-Zero: An Open Source Approach to Scaling Up Reinforcement Learning on the Base Model
- OpenAI o1 System Card
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- 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
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents