boltz.model.models.boltz.BoltzPreview not found

[derby-ding]

When training using confidence.yaml, Error locating target'boltz.model.models.boltz.BoltzPreview'.

If change models.boltz.BoltzPreview to model.Boltz1, we get error Boltz1.init() got an unexpected keyword argument 'run_trunk_and_structure'

[gcorso]

Thanks for pointing this out, I should have fixed the config!
Let me know if you run into any other issues!

[derby-ding]

Thanks for your reply. However, while using the new confidence.ymal to train the confidence layers, there is another error in the 'train.py' says local variable 'file_path' referenced before assignment. So I modified the code to load pretrained checkpoint. Not sure if it's the right way.
if cfg.pretrained and not cfg.resume:
# Load the pretrained weights into the confidence module
model_module = type(model_module).load_from_checkpoint(
cfg.pretrained, strict=False, **(model_module.hparams)
)
if cfg.load_confidence_from_trunk:
checkpoint = torch.load(cfg.pretrained, map_location="cpu")

        # Modify parameter names in the state_dict
        confidence_state_dict = {}
        for key, value in checkpoint["state_dict"].items():
            if not key.startswith("structure_module") and not key.startswith(
                "distogram_module"
            ):
                new_key = "confidence_module." + key
                confidence_state_dict[new_key] = value
        model_module.confidence_module.load_state_dict(confidence_state_dict, strict=False)