Epoch AI releases InnovationEval benchmark; AI agents are far from achieving autonomous research capabilities
Epoch AI uses the new benchmark InnovationEval to test whether Claude Fable 5 and GPT-5.6 Sol can independently invent new training methods, using human-designed SDPO's improvements over GRPO as a reference.
Return to original text
A short introduction is retained here; the full text from the source website is not reposted. For specific wording and subsequent updates, please refer to the original article.
Open AIHOT original