Saving conformer checkpoint and resuming train progress from checkpoint

[LuluW8071]

Confused about saving checkpoint of encoder-decoder models in lightning AI. Is the saving checkpoint technique same for all models?

  checkpoint_callback = ModelCheckpoint(
      monitor='val_loss',
      dirpath="./saved_checkpoint/",
      filename='model-{epoch:02d}-{val_wer:.2f}',
      save_top_k=3,        # 3 Checkpoints
      mode='min'
  )

But how to load the checkpoint again to resume the training? Is the same as normal architecture models like giving ckpt_path in trainer()

  ckpt_path = args.checkpoint_path if args.checkpoint_path else None
  trainer.fit(speech_trainer, data_module, ckpt_path=ckpt_path)   
  trainer.validate(speech_trainer, data_module)

[LuluW8071]

Found the answer the code above is correct