Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2601.17596 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908786163187712 |
|---|---|
| author | Zhang, Yunxiang Zhou, Kang Xu, Zhichao Ramnath, Kiran Zhou, Yun Woo, Sangmin Ding, Haibo Cheong, Lin Lee |
| author_facet | Zhang, Yunxiang Zhou, Kang Xu, Zhichao Ramnath, Kiran Zhou, Yun Woo, Sangmin Ding, Haibo Cheong, Lin Lee |
| contents | Existing machine learning engineering (MLE) agents struggle to iteratively optimize their implemented algorithms for effectiveness. To address this, we introduce MLE-Ideator, a dual-agent framework that separates ideation from implementation. In our system, an implementation agent can request strategic help from a dedicated Ideator. We show this approach is effective in two ways. First, in a training-free setup, our framework significantly outperforms implementation-only agent baselines on MLE-Bench. Second, we demonstrate that the Ideator can be trained with reinforcement learning (RL) to generate more effective ideas. With only 1K training samples from 10 MLE tasks, our RL-trained Qwen3-8B Ideator achieves an 11.5% relative improvement compared to its untrained counterpart and surpasses Claude Sonnet 3.5. These results highlights a promising path toward training strategic AI systems for scientific discovery. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_17596 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Learning to Ideate for Machine Learning Engineering Agents Zhang, Yunxiang Zhou, Kang Xu, Zhichao Ramnath, Kiran Zhou, Yun Woo, Sangmin Ding, Haibo Cheong, Lin Lee Computation and Language Existing machine learning engineering (MLE) agents struggle to iteratively optimize their implemented algorithms for effectiveness. To address this, we introduce MLE-Ideator, a dual-agent framework that separates ideation from implementation. In our system, an implementation agent can request strategic help from a dedicated Ideator. We show this approach is effective in two ways. First, in a training-free setup, our framework significantly outperforms implementation-only agent baselines on MLE-Bench. Second, we demonstrate that the Ideator can be trained with reinforcement learning (RL) to generate more effective ideas. With only 1K training samples from 10 MLE tasks, our RL-trained Qwen3-8B Ideator achieves an 11.5% relative improvement compared to its untrained counterpart and surpasses Claude Sonnet 3.5. These results highlights a promising path toward training strategic AI systems for scientific discovery. |
| title | Learning to Ideate for Machine Learning Engineering Agents |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2601.17596 |