awsm-research / A3Test

16 stars 12 forks source link

Dataset for pretraining #2

Open naresh1205 opened 1 year ago

naresh1205 commented 1 year ago

Hi,

I'm working on a research project and trying to replicate the results in your paper. I'm facing some issues while training. Could you please let me know the English and Code generation pre-training dataset you used? Also, the training script for A3Test has only one major difference than the base PLBart model (one uses PLbart-base and other PLbart-large). Could you please share your training script as I think something might be missing from plbartTraining_Ass.py?