ScienceFlow: A long-horizon agent for ML research, scientific discovery and beyond

2026-08-14

AI Summary

This study addresses the challenges of state discontinuity, difficult dead-end recovery, and inefficient resource allocation in long-horizon autonomous research. We propose an end-to-end autonomous research framework that decomposes tasks into executable workspace segments. Specifically, we introduce the ESTRA mechanism to enable trajectory re-anchoring and state recovery, alongside an evidence-aware controller to optimize resource scheduling. Experimental evaluations on MLE-bench demonstrate that our framework achieves a state-of-the-art performance of 70.22%, surpassing the previous best result by 4.92 percentage points. These findings effectively validate the superiority of the proposed approach in handling complex, long-horizon research tasks, offering a robust solution for maintaining continuity and efficiency in autonomous scientific discovery.

Abstract

Enabling LLM agents to sustain productive, stable, and goal-aligned research over extended horizons is a central challenge for autonomous machine learning and scientific discovery, as progress hinges on continuously managing evolving state, exploration decisions, and computational resources. Pioneering autoresearch agents, despite great success, still lack mechanisms for continuity, recovery from dead ends, and value-driven compute allocation, which inherently undermines overall search efficiency, wastes computational resources, and lowers the chance of ultimate success. To bridge this gap, we introduce ScienceFlow, an end-to-end autoresearch agent framework that organizes long-horizon research work into research segments grounded in executable workspaces. It represents research progress as recoverable executable states, enabling efficient exploration, revision, and execution. Transitions between research segments are governed by Executable-State Transition through Re-Anchoring (ESTRA), which selects either the live state or an archived state as the next anchor and determines whether to continue or redirect the research trajectory. An evidence-aware execution controller allocates resources to physical jobs based on resource availability, remaining budget, and validated progress. We evaluate ScienceFlow on tasks spanning machine learning, scientific modeling, and mathematical optimization. Results on diverse long-horizon benchmarks demonstrate its ability to sustain effective research processes, highlighted by a SOTA 70.22 percent Any-Medal score on the full MLE-bench within a 24-hour budget, outperforming prior reported results by 4.92 percentage points. The efficacy of ScienceFlow further demonstrates that efficient state management, adaptive exploration, and objective-aligned execution are critical for scaling autonomous research beyond short-horizon interactions.

Problem

• Long-horizon agent
• Autonomous research
• State management
• Resource allocation
• Scientific discovery

Innovation

• Long-horizon agent
• Executable state management
• ESTRA
• Evidence-aware execution controller
• Autoresearch framework