You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am trying to replicate your baseline with the pre-trained model provided on the MIMIC dataset.
I got similar F1 but not BLEU. The BLEU score (only on the impression section) I got is 0.04877850514 for CXR-RePaiR-2 (as opposed to 0.069 reported in the paper). Could you share the script for BLEU score or advise me where I could did wrong in the calculation?
Hi,
I am trying to replicate your baseline with the pre-trained model provided on the MIMIC dataset.
I got similar F1 but not BLEU. The BLEU score (only on the impression section) I got is 0.04877850514 for CXR-RePaiR-2 (as opposed to 0.069 reported in the paper). Could you share the script for BLEU score or advise me where I could did wrong in the calculation?
I mostly follow the fast-bleu calculation same in this repository (https://github.com/rajpurkarlab/CXR-Report-Metric).
Thank you so much in advance!
The text was updated successfully, but these errors were encountered: