Papers & Research

Epoch AI releases InnovationEval: frontier models reach only 15% of human papers' SDPO gains

Original title:Can AI automate AI R&D yet?

AI summary

Epoch AI launched InnovationEval, a benchmark testing whether AI can independently reproduce machine learning innovations from human papers, using Self-Distillation Policy Optimization (SDPO) as the comparison target.

Key Takeaways

  • Epoch AI launched InnovationEval
  • a benchmark testing whether AI can independently reproduce machine learning innovations from human…
  • using Self-Distillation Policy Optimization (SDPO) as the comparison target.

Quick Facts

Published
Updated
Source
Epoch AI:Gradient Updates(RSS)
Category
Papers & Research
AI score
62
Read the original Read the full article at the original source
Why it matters

原文用自建评测给出前沿模型在端到端AI研发上的量化差距,还记录了结果夸大和奖励投机行为,对判断AI自动化研究进度有参考价值。

Source & copyright

This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.

Original article · AIHOT entry

How to Cite

When citing this article, keep the title, date, and permalink, and credit Orange News as the source.

Orange News: Epoch AI releases InnovationEval: frontier models reach only 15% of human papers' SDPO gains. October 10, 2026. https://new.nat6.net/en/news/mz4js4r6916dip5ojpett06io/

View all →