Tag
提出了 MLExpert，一个 35B 混合专家（MoE）智能体模型，通过分层监督微调和窗口化交互强化学习（WIRL）自主优化长周期机器学习工程工作流，在 MLE-Bench Lite 上将基模型奖牌率从 37.9% 提升至 72.7%，媲美更大的前沿模型。
This paper introduces Meta-Skill, a method letting a frozen Builder model learn reusable principles from Target execution feedback to construct better agent harnesses for unseen tasks, improving performance by 8.95 points on Harness-Bench and NewtonBench. It suggests a path toward system-level self-improvement by learning to build better environments rather than changing model weights.
Awesome AI4AI is a living map that tracks AI improving AI, including latest papers, news, rankings, and benchmarks, updated weekly to serve as a dynamic resource for the field.
This paper introduces OpenMLE, an open full-stack system for studying recursive self-improvement in machine learning engineering, and presents Frontis-MA1, a 35B model post-trained as a meta-evolution agent. It shows significant improvement over its base model on MLE-BenchLite and transfers to held-out benchmarks, with weights and code released.
XYZ AI Lab releases XYZ-Aquila-mini, an open-weight thinking model for agentic deep search, fine-tuned from Qwen3.6-35B-A3B, achieving top benchmark scores among sub-40B open-weight models.
XYZ AI Lab releases XYZ-Aquila-pro, an open-weight thinking model for agentic deep search, post-trained from Qwen3.5-397B-A17B via a bounded-exploration AI4AI pipeline, with strong benchmark results in sub-400B open-weight comparisons.