How can I resume training pl.Trainer after interruption?

[andreys42]

I have Model and Trainer pytorch-lightning objects which are initialized as follows:

model = nn.DataParallel (FaultNetPL(batch_size=5)).cuda()
trainer = Trainer( checkpoint_callback=checkpoint_callback, 
                  max_epochs=500,gpus=1,
                  logger=logger)

Then I started training using:

trainer.fit(model)

But training was interrupted and now I would like to resume it using checkpoint from N-th iteration So i tried initialize model and trainer as:

model = FaultNetPL.load_from_checkpoint('experiments/VNET/epoch=77-avg_valid_iou=0.7604.ckpt',batch_size=5)
trainer = Trainer(resume_from_checkpoint = 'epoch=77-avg_valid_iou=0.7604.ckpt', 
                  checkpoint_callback=checkpoint_callback, 
                  max_epochs=500,gpus=1,
                  logger=logger)

But again and again training starting from scratch (from 0th epoch and huge error). What did I miss?