How can I use the finetuned model to eval

[Shuweis]

I finetuned the model animate-anything on other datasets and save the checkpoint in the new folder file. However, I saw the eval process is also load the pretrained model (animate_anything_512_v102) of the config.yaml in the new folder file. How can I eval my fine-tuned model to get the new results?

[sculmh]

You just need to change the pretrained_model_path in the YAML file to your new folder path. The new config.yaml simply copies the configuration you used during fine-tuning.