Closed CiaraG98 closed 2 years ago
Model is actually on the checkpoint itself (runs folder), you need to rename last checkpoint.pt to pytorch_model.bin then load model as in normal huggingface models (from_pretrained). There seems to be a bug in last checkpoint rename.
Where exactly is the model saved after executing train.py? I am training on a different laptop and want to run interact.py on my laptop with the model I trained on another laptop.