Synthetic Sandbox for Training Machine Learning Engineering Agents

• URL: http://arxiv.org/abs/2604.04872v1
• Date: Mon, 06 Apr 2026 17:19:29 GMT
• Title: Synthetic Sandbox for Training Machine Learning Engineering Agents
• Abstract summary: SandMLE is a multi-agent framework that generates verifiable synthetic MLE environments from a small number of seed tasks.<n>We show that SandMLE reduces execution time by over 13 times, enabling large-scale, on-policy trajectory-wise RL for the first time in the MLE domain.
• Score: 26.34347159807133
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: As large language model agents advance beyond software engineering (SWE) tasks toward machine learning engineering (MLE), verifying agent behavior becomes orders of magnitude more expensive: while SWE tasks can be verified via fast-executing unit tests, MLE verification requires running full ML pipelines -- data preprocessing, model training, and metric evaluation -- on large datasets at each rollout step, rendering trajectory-wise on-policy reinforcement learning (RL) prohibitively slow. Existing approaches retreat to supervised fine-tuning (SFT) or offline proxy rewards, sacrificing the exploration and generalization benefits of on-policy RL. We observe that sandbox data size is the primary source of this bottleneck. Based on this insight, we introduce SandMLE, a multi-agent framework that generates diverse, verifiable synthetic MLE environments from a small number of seed tasks, preserving the structural and technical complexity of real-world problems while constraining datasets to micro-scale (each task is paired with only 50-200 training samples). Through extensive experiments, we show that SandMLE reduces execution time by over 13 times, enabling large-scale, on-policy trajectory-wise RL for the first time in the MLE domain. On MLE-bench-lite, SandMLE yields significant gains over SFT baselines across Qwen3-8B, 14B, and 30B-A3B, with relative medal rate improvements ranging from 20.3% to 66.9%. Furthermore, the trained policy generalizes across unseen agentic scaffolds, achieving up to 32.4% better HumanRank score on MLE-Dojo.

Related papers

• Guiding LLM Post-training Data Engineering with Model Internals from Sparse Autoencoders [49.87696007629591]
  Post-training data engineering largely relies on external signals and ignores rich intrinsic signals lying in model internals.<n>We propose SAERL, a data engineering framework for reinforcement learning (RL)<n>It models three intrinsic data properties: diversity, difficulty, and quality, using model internals extracted with Sparse Autoencoder (SAE), an advanced mechanistic interpretability tool.
  arXiv  Detail & Related papers  (2026-05-26T17:55:59Z)
• AceGRPO: Adaptive Curriculum Enhanced Group Relative Policy Optimization for Autonomous Machine Learning Engineering [52.67783579040657]
  AceGRPO is a machine learning system that prioritizes tasks at the agent's learning frontier to maximize learning efficiency.<n>Our trained Ace-30B model achieves a 100% valid submission rate on MLE-Bench-Lite, approaches the performance of proprietary frontier models, and outperforms larger open-source baselines.
  arXiv  Detail & Related papers  (2026-02-08T10:55:03Z)
• From Self-Evolving Synthetic Data to Verifiable-Reward RL: Post-Training Multi-turn Interactive Tool-Using Agents [23.583947864141162]
  EigenData is a hierarchical multi-agent engine that synthesizes tool-grounded dialogues together with executable per-instance checkers.<n>Building on the synthetic data, we develop an RL recipe that first fine-tunes the user model and then applies GRPO-style training.<n>Our results suggest a scalable pathway for bootstrapping complex tool-using behaviors without expensive human annotation.
  arXiv  Detail & Related papers  (2026-01-30T06:01:23Z)
• Skywork-SWE: Unveiling Data Scaling Laws for Software Engineering in LLMs [19.766885088032932]
  Software engineering (SWE) has emerged as a crucial testbed for next-generation LLM agents.<n>Most existing datasets are limited to only a few thousand GitHub-sourced instances.<n>We propose an incremental, automated data-curation pipeline that systematically scales both the volume and diversity of SWE datasets.
  arXiv  Detail & Related papers  (2025-06-24T03:53:36Z)
• Ring-lite: Scalable Reasoning via C3PO-Stabilized Reinforcement Learning for LLMs [51.21041884010009]
  Ring-lite is a Mixture-of-Experts (MoE)-based large language model optimized via reinforcement learning (RL)<n>Our approach matches the performance of state-of-the-art (SOTA) small-scale reasoning models on challenging benchmarks.
  arXiv  Detail & Related papers  (2025-06-17T17:12:34Z)
• First SFT, Second RL, Third UPT: Continual Improving Multi-Modal LLM Reasoning via Unsupervised Post-Training [37.80193099472551]
  We propose MM-UPT, a simple yet effective framework for unsupervised post-training of MLLMs.<n>Our experiments demonstrate that such training method effectively improves the reasoning ability of Qwen2.5-VL-7B.<n>We extend our framework to a data self-generation setting, designing two strategies that prompt the MLLM to synthesize new training samples.
  arXiv  Detail & Related papers  (2025-05-28T15:11:16Z)
• LaMDAgent: An Autonomous Framework for Post-Training Pipeline Optimization via LLM Agents [3.6117068575553595]
  We introduce LaMDAgent, a framework that autonomously constructs and optimize full post-training pipelines.<n>LaMDAgent improves tool-use accuracy by 9.0 points while preserving instruction-following capabilities.<n>It uncovers effective post-training strategies that are often overlooked by conventional human-driven exploration.
  arXiv  Detail & Related papers  (2025-05-28T04:30:51Z)
• TL-Training: A Task-Feature-Based Framework for Training Large Language Models in Tool Use [72.32614703504122]
  Large language models (LLMs) achieve remarkable advancements by leveraging tools to interact with environments.<n>Standard supervised fine-tuning approach, which relies on large-scale datasets, often overlooks task-specific characteristics in tool use.<n>We propose TL-Training, a task-feature-based framework that mitigates the effects of suboptimal training data.
  arXiv  Detail & Related papers  (2024-12-20T02:21:36Z)
• MALT: Improving Reasoning with Multi-Agent LLM Training [67.76186488361685]
  MALT (Multi-Agent LLM Training) is a novel post-training strategy that divides the reasoning process into generation, verification, and refinement steps.<n>On MATH, GSM8K, and CSQA, MALT surpasses the same baseline LLM with a relative improvement of 15.66%, 7.42%, and 9.40% respectively.
  arXiv  Detail & Related papers  (2024-12-02T19:30:36Z)
• PLANRL: A Motion Planning and Imitation Learning Framework to Bootstrap Reinforcement Learning [13.564676246832544]
  We introduce PLANRL, a framework that chooses when the robot should use classical motion planning and when it should learn a policy. PLANRL switches between two modes of operation: reaching a waypoint using classical techniques when away from the objects and fine-grained manipulation control when about to interact with objects. We evaluate our approach across multiple challenging simulation environments and real-world tasks, demonstrating superior performance in terms of adaptability, efficiency, and generalization compared to existing methods.
  arXiv  Detail & Related papers  (2024-08-07T19:30:08Z)
• Multitask Adaptation by Retrospective Exploration with Learned World Models [77.34726150561087]
  We propose a meta-learned addressing model called RAMa that provides training samples for the MBRL agent taken from task-agnostic storage. The model is trained to maximize the expected agent's performance by selecting promising trajectories solving prior tasks from the storage.
  arXiv  Detail & Related papers  (2021-10-25T20:02:57Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.