NomicBertModel.forward() got an unexpected keyword argument 'return_dict'

#10
by edsealing - opened

Using SentenceTransformer to load the model:

  model_name = 'nomic-ai/nomic-embed-text-v1'
  embedmodel = SentenceTransformer(model_name, trust_remote_code=True)
  return setup_model_device(embedmodel, model_name)

When attempting to call the embedding function, I receive an error.

  def compute(self, docs: list[str]) -> list[Optional[Item]]:
    """Call the embedding function."""
    prefixed_docs = [f"search_document {doc}" for doc in docs]
    return chunked_compute_embedding(
      self._model.encode, prefixed_docs, self.local_batch_size * 16, chunker=clustering_spacy_chunker
    )

NomicBertModel.forward() got an unexpected keyword argument 'return_dict'

Nomic AI org

thanks for the headsup!

Just fixed and tested

from sentence_transformers import SentenceTransformer
model_name = 'nomic-ai/nomic-embed-text-v1'
embedmodel = SentenceTransformer(model_name, trust_remote_code=True)

test = ["search_query: who does shohei ohtani play for?", "search_document: shohei ohtani plays for the los angeles dodgers"]
embedmodel.encode(test)

Let me know if it doesn't work on your end

That did the trick! Fastest fix I think I've ever seen...

Cheers!

edsealing changed discussion status to closed

Sign up or log in to comment