ValueError

#1
by CyunNaam - opened

When importing your model through the transformer library, an error message is displayed: "ValueError: Unrecognized model in Edoardo-BS/HuBERT-ECG. Should have a model_type key in its config.json, or contain one of the following strings in its name".

Hi @CyunNaam !
Without a code snippet to reproduce the error you're encountering it can be hard to provide a good and precise aswer, but I think I see your point.
Altough relying on HuBERT configuration and HuBERT model, we haven't set our own model_type yet. We'll do it as soon as possibile, enabling the possibility to load the model as an off-the-shelf transformers resource, but for now we suggest you to simply use torch.load.
Please, consider taking a look at our code repo (e.g. finetune.py)

Should you have any suggestion for possible improvements, please let us know.

Happy coding!

Hi @CyunNaam !
Without a code snippet to reproduce the error you're encountering it can be hard to provide a good and precise aswer, but I think I see your point.
Altough relying on HuBERT configuration and HuBERT model, we haven't set our own model_type yet. We'll do it as soon as possibile, enabling the possibility to load the model as an off-the-shelf transformers resource, but for now we suggest you to simply use torch.load.
Please, consider taking a look at our code repo (e.g. finetune.py)

Should you have any suggestion for possible improvements, please let us know.

Happy coding!

Thank you very much for your answer.

This comment has been hidden
Edoardo-BS changed discussion status to closed
This comment has been hidden
Edoardo-BS changed discussion status to open
Edoardo-BS changed discussion status to closed

Hi @CyunNaam !
Without a code snippet to reproduce the error you're encountering it can be hard to provide a good and precise aswer, but I think I see your point.
Altough relying on HuBERT configuration and HuBERT model, we haven't set our own model_type yet. We'll do it as soon as possibile, enabling the possibility to load the model as an off-the-shelf transformers resource, but for now we suggest you to simply use torch.load.
Please, consider taking a look at our code repo (e.g. finetune.py)

Should you have any suggestion for possible improvements, please let us know.

Happy coding!

Thank you very much for your answer.

Hello again @CyunNaam ! Now there's the possibility to load HuBERT-ECG pre-trained backbone using AutoModel.from_pretrained.
Please take a look at the usage section in the model card.
Thank you!

Edoardo-BS changed discussion status to open
Edoardo-BS changed discussion status to closed

Sign up or log in to comment