Clarification around datasets

[elangovana]

Hi,
Which eval set are you using for Yelp, Amazon and semeval?

1. https://huggingface.co/datasets/yelp_polarity/tree/main Are you using the test set here which has 38k samples
2. Amazon polarity - Are u using this one https://huggingface.co/datasets/amazon_polarity, which has 400,000 test samples?
3. For sem eval which dev set are you using
4. Also for the Orig. & Rev. (3.4k) revised train set (referred in Table 5 and 6) with counterfactual , is this the training data https://raw.githubusercontent.com/acmi-lab/counterfactually-augmented-data/master/sentiment/combined/paired/train_paired.tsv