AutoMLGen: Navigating Fine-Grained Optimization for Coding Agents
This addresses the need for more efficient and effective coding agents in ML engineering scenarios like competitions, though it appears incremental by building on existing methods like MCTS.
The paper tackles the problem of LLMs lacking fine-grained domain priors and limited knowledge transfer in Machine Learning Engineering tasks like AutoML, introducing AutoMLGen which integrates a domain knowledge base and Monte Carlo Graph Search to achieve state-of-the-art performance on MLE-Bench, such as improved average medal and valid submission rates under a reduced 12-hour budget.
Large language models (LLMs) have shown impressive performance in general programming tasks. However, in Machine Learning Engineering (MLE) scenarios such as AutoML and Kaggle competitions, achieving high performance depends heavily on expert intervention and repeated adjustments rather than simply generating correct code. When applied directly to these tasks, LLMs often lack fine-grained domain priors, and existing MLE approaches that use linear or tree-structured searches limit knowledge transfer to adjacent hierarchical links. As a result, they cannot leverage past full trajectories or share information across branches, limiting self-evolving ability and search space diversity. To address these limitations, we introduce AutoMLGen, an LLM-based coding agent that integrates a domain knowledge base for high-quality prior guidance and Monte Carlo Graph Search (MCGS) for efficient exploration. MCGS retains the tree-guided exploration of MCTS while embedding a graph structure into the expansion stage to enable dynamic path reorganization, historical trajectory reuse, and multi-solution fusion to support both self-evolution and collaborative learning. Combined with fine-grained operator sets, this design improves stability and accelerates convergence. Evaluation on the MLE-Bench shows that AutoMLGen achieves state-of-the-art performance in numerous dimensions, such as the average medal rate and the valid submission rate, under a 12-hour budget (half the standard runtime). The code is available at https://github.com/Alpha-Innovator/InternAgent.