Actually, I’m implementing a multi-task learning model, using the trainer class in huggingface. Since huggingface trainer as well as dataloader except for one model at a go, thus I needed to modify both the classes, Now I’m just confused, about how to evaluate the running model while the model is training.
Note I’m using transformers version 2.11.0
I tried using the evaluation_strategy under trainerarguments but it’s giving some errors.
Also here is the link to a reference colab notebook here
Can someone help me with this? thankyou!