Closed omihub777 closed 2 months ago
Hello, thank you for the issue!
Your guess is right. We convert labels to integer with the following dictionary: https://huggingface.co/datasets/sbintuitions/JMTEB/blob/7e8ad98b751ab42c572ceb4e7723a2885a3d2cfe/reranking.py#L24
Thanks!
Thank you for your great work. I have a quick question regarding
relevance_scores
in ESCI. In your version of ESCI dataset, therelevance_scores
takes one of the values from 0 to 3. What does each value correspond to the original labels (i.e. Exact / Substitute / Complement / Irrelevant)? My guess would be0: Irrelevant
,1: Complement
,2: Substitute
,3: Exact
, but want your confirmation. Thanks.