THUDM / P-tuning-v2

An optimized deep prompt tuning strategy comparable to fine-tuning across scales and tasks
Apache License 2.0
1.96k stars 198 forks source link

ReCoRD dataset #56

Open markus583 opened 1 year ago

markus583 commented 1 year ago

It is not clear how you deal with the ReCoRD dataset - the collater, metrics, and so on. In your final version, do you use the SuperGlueDatasetForRecord or just the regular SuperGlueDataset? It is really unclear.

If have tried both and also tried to fix some of the errors involved but have not found a way to work it out. It would be highly appreciated and very helpful if you could specify this a bit. Thanks!