Trying to restore training state but checkpoint contains only the model. This is probably due to `ModelCheckpoint.save_weights_only` being set to `True`

1

HI, I am using Pytorch Lightning, trying to restore a model, I have de model_epoch=15.ckpt file and would like to restore from here, so I introduced the resume_from_checkpoint in the trainer, but I get the following error:
Trying to restore training state but checkpoint contains only the model. This is probably due to ModelCheckpoint.save_weights_only being set to True.

This is because I put save_weights_only=True in the ModelCheckpoint.
What should I do to restore the model?

Here is my code:

def main():
    exp_folder = get_experiment_folder()
    train_dataset = Hotelimages(split='train')
    valid_dataset = Hotelimages(split='valid')
    batch_size = 32 
    train_dl = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)
    valid_dl = DataLoader(valid_dataset, batch_size=batch_size, shuffle=True)
    model = SwinTransformerFineTuning()
    model_checkpoint = ModelCheckpoint(exp_folder, save_last=True, filename='model_{epoch}',
                                       save_weights_only=True, every_n_epochs=5)
    trainer = pl.Trainer(precision=16, default_root_dir=exp_folder, callbacks=[model_checkpoint], max_epochs=100, accelerator='gpu', devices=1,
                         resume_from_checkpoint=run_folder/'1/model_epoch=14.ckpt')
    trainer.fit(model, train_dl, valid_dl)
2

I already had this problem. You can solved it this way.

model = SwinTransformerFineTuning.load_from_checkpoint(
    # hparams = ..., # if needed 
    checkpoint_path = run_folder/'1/model_epoch=14.ckpt'
)
model.train()
# ...

trainer = pl.Trainer(resume_from_checkpoint=None)
trainer.fit(model, train_dl, valid_dl)