artidoro / frank

FRANK: Factuality Evaluation Benchmark
MIT License
52 stars 4 forks source link

BERTscore implementation #3

Closed Lukecn1 closed 3 years ago

Lukecn1 commented 3 years ago

Hi Artidoro,

Can you provide the details on the BERTscore implementation you have used to derive the scores in the data?

I am having difficulty replicating them exactly.

artidoro commented 3 years ago

Hello, I just verified and the results in the baseline_factuality_metrics_outputs.json are incorrect. I apologize for the issue and will update the file and upload the latest results to the website.

Thank you very much for reporting the problem!

Lukecn1 commented 3 years ago

I just reran my scores and all looks good now, thanks for the quick update :)