ML Engineering
NatureBench
Published Match-SOTA outcomes for 23 coding-agent configurations on 90 scientific ML tasks, with complete available native run files and recorded validity judgments.
181items
23subjects
CC-BY-4.0 for the released process traces and original result metadata; MIT for original task/code contributions. Task-specific third-party data retain their own licenses and are not mirrored here.license
sciencedomain
ml_engineeringdomain
textmodality
item-level responses released
Saturation status: No
Response matrix
Loading response matrix…
Correct (1)Incorrect (0)Unobserved
Scale: 1 = correct · 0 = incorrect