Example for Using GPU

Hi Folks, I've been trying to get this running on a g4dn.2xlarge on EC2 for hours with zero success. Everytime I load the model via from_pretrained, it looks like defaults to CPU. I've tried passing in parameters like so: ``` relik_model = Relik.from_pretrained("sapienzanlp/relik-entity-linking-small",device="cuda",overrides={"device":"cuda"}) relik_relationships = Relik.from_pretrained("sapienzanlp/relik-relation-extraction-nyt-large",device="cuda",overrides={"device":"cuda"}) ``` But that doesn't seem to work either: I still see: `[relik.retriever.indexers.base.from_pretrained:485] [PID:1605] [RANK:0] { '_target_': 'relik.retriever.indexers.inmemory.InMemoryDocumentIndex', 'device': 'cpu', 'metadata_fields': ['definition'], 'name_or_path': None, 'precision': None, 'separator': ' <def> ', 'use_faiss': False, }` I'm using Amazon's Deep Learning AMI for Ubuntu. Any advice would be appreciated -- I'm a bit of a newb. Thanks!