FML-Bench shows a simple greedy hill-climber nearly matches tree search on dense-opportunity tasks while an adaptive agent that broadens search on stagnation outperforms six baselines across 18 tasks.
Mars: Modular agent with reflective search for automated ai research
12 Pith papers cite this work. Polarity classification is still indexing.
12
Pith papers citing it
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.
years
2026 12representative citing papers
Arbor combines a coordinator, executors, and a hypothesis tree to enable cumulative autonomous research, outperforming Codex and Claude Code by over 2.5x on six real tasks and reaching 86.36% Any Medal on MLE-Bench Lite.
DrugSAGE accumulates cross-task memory of skills, statistical evidence, and recurring errors to let LLM agents achieve top-ranked performance on molecular property prediction tasks with reduced or zero test-time search.
DAgger-style training with turn-level policy interpolation raises 4B and 8B LLM agents to 27.3% and 29.8% on SWE-bench Verified, beating several larger published systems.
AIBuildAI uses a manager agent and three LLM sub-agents to fully automate AI model development and achieves a 63.1% medal rate on MLE-Bench, matching experienced human engineers.
HypoExplore uses LLMs for hypothesis-driven evolutionary search with a Trajectory Tree and Hypothesis Memory Bank to discover lightweight vision architectures, reaching 94.11% accuracy on CIFAR-10 from an 18.91% baseline and generalizing to other datasets including state-of-the-art on MedMNIST.
AIRA₂ improves AI research agents via asynchronous multi-GPU workers, hidden consistent evaluation, and interactive ReAct agents, reaching 81.5-83.1% percentile rank on MLE-bench-30 and exceeding human SOTA on 6 of 20 AIRS-Bench tasks.
EurekAgent achieves new state-of-the-art results on mathematics, kernel engineering, and machine learning tasks by engineering agent environments for autonomous scientific discovery, including a 26-circle packing result at under $11 API cost.
CBR integration into R&D-Agent with Gemma 4 31B yields directionally higher accuracy and lower variance than baseline on one of two Kaggle competitions.
AIBuildAI-2 introduces a knowledge-enhanced agent with a hierarchical evolving external knowledge base that dynamically loads relevant AI development expertise, achieving first place on MLE-Bench at 70.7% medal rate.
MLEvolve is a self-evolving multi-agent LLM system with Progressive MCGS, Retrospective Memory, and adaptive coding modes that reports SOTA medal and submission rates on MLE-Bench under a 12-hour budget while outperforming AlphaEvolve on math tasks.
citing papers explorer
-
AIRA_2: Overcoming Bottlenecks in AI Research Agents
AIRA₂ improves AI research agents via asynchronous multi-GPU workers, hidden consistent evaluation, and interactive ReAct agents, reaching 81.5-83.1% percentile rank on MLE-bench-30 and exceeding human SOTA on 6 of 20 AIRS-Bench tasks.