FrogNano: Training a 4B Coding Agent via Online Task Synthesis
cs.AI
Submitted: 2026-09-07
Updated: 2026-10-05
Code: https://github.com/THUDM/slime
Project page: https://microsoft.github.io/debug-gym
License: http://creativecommons.org/licenses/by/4.0/
The gist: We present FrogNano, a 4B coding agent designed to tackle software engineering (SWE) tasks efficiently and effectively, even under resource-constrained environments.
Terminology
Abstract
We present FrogNano, a 4B coding agent designed to tackle software engineering (SWE) tasks efficiently and effectively, even under resource-constrained environments. It is post-trained exclusively via RL on around 1,500 SWE environments with synthetic tasks. A key ingredient for improving performance is an online task synthesis pipeline that creates tasks calibrated to the frontier of learnability for the current checkpoint. This report provides evidence that competitive small coding agents can be trained with synthetic tasks alone, without traditional distillation from larger models, and that generating tasks at the learnability frontier of the current agent is important. We report details on the training methodology, evaluations across diverse environments, and in-depth analyses, serving as a foundation for our ongoing exploration of lightweight yet capable coding agents that can run on minimal hardware.
Sources
- How Do AI Agents Spend Your Money? Analyzing and Predicting Token Consumption in Agentic Coding Tasks
- Agent-RLVR: Training Software Engineering Agents via Guidance and Environment Rewards
- SWE-MeM: Learning Adaptive Memory Management for Long-Horizon Coding Agents
- Training Long-Context, Multi-Turn Software Engineering Agents with Reinforcement Learning
- Single-Rollout Asynchronous Optimization for Agentic Reinforcement Learning
- DORA: A Scalable Asynchronous Reinforcement Learning System for Language Model Training
- Qwen2.5-Coder Technical Report
- Tmax: A simple recipe for terminal agents
- R2E-Gym: Procedural Environments and Hybrid Verifiers for Scaling Open-Weights SWE Agents
- Recursive Synthesis for Long-Horizon Terminal Tasks
- SPADE: Self-Play in Adaptive Synthetic Executable Environments
- SCALECUA: Scaling Computer Use Agents with Verifiable Task Synthesis and Efficient Online RL
- FailForge: Distilling Procedural Competence from Persistent Failures into Code Agents
- CalibForge: Adversarial Solver Calibration for Scaling Learnable Terminal Tasks
- Training Software Engineering Agents and Verifiers with SWE-Gym
- Orchard: An Open-Source Agentic Modeling Framework
- PipelineRL: Faster On-policy Reinforcement Learning for Long Sequence Generation
- Rethinking the Trust Region in LLM Reinforcement Learning
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- BugPilot: Complex Bug Generation for Efficient Learning of SWE Skills
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