Serveless memory problem when deploy Wav2Vec2 with custom inference code

[philschmid]

Are you providing the model via s3 uri or hub configuration. You have to provide it via s3://path as model_data since the hub configuration is not loading the kenlm model.