academic-rhetoric-lab / external_error_analysis.csv
Hugdrigo's picture
Upload 27 files
be1435c verified
Raw History Blame Contribute Delete
475 Bytes
label,support,recall,diagnosis
background,5,0.0,Model does not transfer background cues reliably to biomedical abstracts.
objective,5,0.4,Objective signals transfer partially.
method,6,0.833,Most stable transferred class in the smoke test.
result,6,0.333,Result cues are frequently confused under domain shift.
implication,6,0.167,PubMed conclusions do not align cleanly with the broader implication label.
limitation,0,,No direct PubMed RCT source label counterpart.