MARS: Modular Agent with Reflective Search for Automated AI Research

TL;DR

MARS is an autonomous AI research framework that combines budget-aware planning, modular design, and reflective memory to optimize complex machine learning tasks efficiently.

Contribution

It introduces a novel modular agent with reflective search that balances performance and cost, advancing automated AI research capabilities.

Findings

Achieves state-of-the-art results on MLE-Bench among open-source frameworks.

Demonstrates 63% of insights originate from cross-branch transfer, showing effective generalization.

Maintains competitiveness with top global leaderboard methods.

Abstract

A critical bottleneck in automating AI research is the execution of complex machine learning engineering (MLE) tasks. MLE differs from general software engineering due to computationally expensive evaluation (e.g., model training) and opaque performance attribution. Current LLM-based agents struggle here, often generating monolithic scripts that ignore execution costs and causal factors. We introduce MARS (Modular Agent with Reflective Search), a framework optimized for autonomous AI research. MARS relies on three pillars: (1) Budget-Aware Planning via cost-constrained Monte Carlo Tree Search (MCTS) to explicitly balance performance with execution expense; (2) Modular Construction, employing a "Design-Decompose-Implement" pipeline to manage complex research repositories; and (3) Comparative Reflective Memory, which addresses credit assignment by analyzing solution differences to distill high-signal insights. MARS achieves state-of-the-art performance among open-source frameworks on MLE-Bench under comparable settings, maintaining competitiveness with the global leaderboard's top methods. Furthermore, the system exhibits qualitative "Aha!" moments, where 63% of all utilized lessons originate from cross-branch transfer, demonstrating that the agent effectively generalizes insights across search paths.