facebook/tts_transformer-fr-cv7_css10 · Errors with the usage of model card

Hello,
Using the model card usage code, as is, to run the model, I encountered some issues.

The first one is TypeError: 'TTSTransformerModel' object is not subscriptable which can be resolved adding brackets around "model" in line
generator = task.build_generator(model, cfg)
Fix : generator = task.build_generator([model], cfg)
(Thanks to https://stackoverflow.com/questions/71029685/not-able-to-execute-sample-code-provided-in-hugging-faces-models-card)

Then, I encountered an issue because of model weights are cuda.FloatTensor and input is just FloatTensor.
RuntimeError: Input type (torch.FloatTensor) and weight type (torch.cuda.FloatTensor) should be the same or input should be a MKLDNN tensor and weight is a dense tensor
Would it be possible to precise how to either use GPU or CPU to allow quick getting started with this model ?
Thanks ! :)