AutoResearchExam: Measuring agents' ability to improve and generalize

What happened

AutoResearchExam measures agents' ability to improve and generalize on open ended ML research tasks.

Summary assembled by rule from the sources below

Sources