Combee: Scaling Prompt Learning for Self-Improving Language Model Agents
cs.AI, cs.CL, cs.LG
Submitted: 2026-04-05
Updated: 2026-09-25
Comments: COLM 2026
Code: https://github.com/gepa-ai/gepa
License: http://creativecommons.org/licenses/by/4.0/
The gist: Recent advances in prompt learning allow large language model agents to acquire task-relevant knowledge from inference-time context without parameter changes.
Terminology
Abstract
Recent advances in prompt learning allow large language model agents to acquire task-relevant knowledge from inference-time context without parameter changes. For example, existing methods (like ACE or GEPA) can learn system prompts to improve accuracy based on previous agent runs. However, these methods primarily focus on single-agent or low-parallelism settings. This fundamentally limits their ability to efficiently learn from a large set of collected agentic traces. It would be efficient and beneficial to run prompt learning in parallel to accommodate the growing trend of learning from many agentic traces or parallel agent executions. Yet without a principled strategy for scaling, current methods suffer from quality degradation with high parallelism. To improve both the efficiency and quality of prompt learning, we propose Combee, a novel framework to scale parallel prompt learning for self-improving agents. Combee speeds up learning and enables running many agents in parallel while learning from their aggregate traces without quality degradation. To achieve this, Combee leverages parallel scans and employs an augmented shuffle mechanism; Combee also introduces a dynamic batch size controller to balance quality and delay. Evaluations on AppWorld, Terminal-Bench, Formula, and FiNER demonstrate that Combee achieves up to 17x speedup over previous methods with comparable or better accuracy and equivalent cost.
Sources
- GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning
- CL-bench: A Benchmark for Context Learning
- Learning to Share: Selective Memory for Efficient Parallel Agentic Systems
- Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour
- SWE-bench: Can Language Models Resolve Real-World GitHub Issues?
- More Agents Is All You Need
- FiNER: Financial Numeric Entity Recognition for XBRL Tagging
- MathVista: Evaluating Mathematical Reasoning of Foundation Models in Visual Contexts
- FrontierCS: Evolving Challenges for Evolving Intelligence
- An Empirical Model of Large-Batch Training
- Terminal-Bench: Benchmarking Agents on Hard, Realistic Tasks in Command Line Interfaces
- ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory
- Scaling Large Language Model-based Multi-Agent Collaboration
- Don't Decay the Learning Rate, Increase the Batch Size
- Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters
- Dynamic Cheatsheet: Test-Time Learning with Adaptive Memory
- AppWorld: A Controllable World of Apps and People for Benchmarking Interactive Coding Agents
- FinLoRA: Benchmarking LoRA Methods for Fine-Tuning LLMs on Financial Datasets
- Voyager: An Open-Ended Embodied Agent with Large Language Models
- Mixture-of-Agents Enhances Large Language Model Capabilities
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