Load model weight local

[ndlongvn]

I have already downloaded a checkpoint of the model from Hugging Face. How can I use these weights for model inference?

[avidale]

@ndlongvn If you are using the HuggingFace version of the weights, then you can refer to the documentation of the HF re-implementation of Seamless for the inference recipes. You will need only to replace the model names in their code (such as "facebook/hf-seamless-m4t-medium") with the path to your local directory with the checkpoint.