Science
Agentic Review Systems — Perturbation Benchmark
Released early per-error detection results for a Claude Opus 4.6 review workflow on six perturbed papers. Items contain the released paper text; grading criteria identify the injected errors. This is a partial experimental slice, not the full paper evaluation.
105items
3subjects
MITlicense
sciencedomain
reasoningdomain
textmodality
item-level responses released
Saturation status: No
Response matrix
Loading response matrix…
Correct (1)Incorrect (0)Unobserved
Scale: 1 = correct · 0 = incorrect