Evolving Memory, Learning to Meta-Control: Test-Time Adaptation for Experience-augmented LLM Agents
cs.LG, cs.AI, cs.CL
Submitted: 2026-11-01
Updated: 2026-11-01
Code: https://github.com/bigcode-project/bigcodebench
Terminology
Sources
- Experiential Reflective Learning for Self-Improving LLM Agents
- Recursive Self-Improvement in AI: From Bounded Self-Refinement to Autonomous Research Loops
- SWE-Exp: Experience-Driven Software Issue Resolution
- Rational Metareasoning for Large Language Models
- Why AI systems don't learn and what to do about it: Lessons on autonomous learning from cognitive science
- A Comprehensive Survey of Self-Evolving AI Agents: A New Paradigm Bridging Foundation Models and Lifelong Agentic Systems
- Memp: Exploring Agent Procedural Memory
- A Survey of Self-Evolving Agents: What, When, How, and Where to Evolve on the Path to Artificial Super Intelligence
- SWE-bench: Can Language Models Resolve Real-World GitHub Issues?
- Causal Reasoning and Large Language Models: Opening a New Frontier for Causality
- Adaptive Tool Use in Large Language Models with Meta-Cognition Trigger
- DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models
- LLM-Powered Hierarchical Language Agent for Real-time Human-AI Coordination
- CLIN: A Continually Learning Language Agent for Rapid Task Adaptation and Generalization
- ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory
- Cognitive Demand Steering for Adaptive Meta-Reasoning in Large Language Models
- Causal Language Modeling Can Elicit Search and Reasoning Capabilities on Logic Puzzles
- Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters
- Agent KB: Leveraging Cross-Domain Experience for Agentic Problem Solving
- Kimi K2.5: Visual Agentic Intelligence
Related papers
- A Physics-Regulated Neural Framework for Learning 3D Grain Growth Dynamics
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis