nomic-ai

/

nomic-embed-text-v1.5

like 359

I got it working using txtai:

from txtai import Embeddings

# Create embeddings instance with a semantic graph
embeddings = Embeddings({
  "autoid": "uuid5",
  "path": "nomic-ai/nomic-embed-text-v1.5",
  "instructions": {
    "query": "query: ",
    "data": "passage: "
  },
  "content": True,
  "graph": {
      "approximate": False,
      "topics": {}
  }
})

# Load dataset
wikipedia = Embeddings()
wikipedia.load(provider="huggingface-hub", container="neuml/txtai-wikipedia")

query = """
SELECT id, text FROM txtai
order by percentile desc
LIMIT 1000
"""

embeddings.index(wikipedia.search(query))
print(embeddings.search("select id, article, score from txtai where similar(:x)", parameters={"x": "operating system"}))

Two issues...

1. It uses remote code. So, make sure you press 'y' to accept when you run the above script or make changes to this file:

txtai/models/models.py line 204:

return models[task](path,trust_remote_code=True) if task in models else path

2. For some weird reason, it expects the "nomic-ai/nomic-embed-text-v1.5/model.safetensors" file in path where the above script is run. I simply run the following command where the script is:

   huggingface-cli download nomic-ai/nomic-embed-text-v1.5 --local-dir ./nomic-ai/nomic-embed-text-v1.5

This is coming from https://huggingface.co/nomic-ai/nomic-embed-text-v1-unsupervised/blob/62fe27d25832db69e9002f0ba71f9b3c2e7bad63/modeling_hf_nomic_bert.py#L52

Referenced here:

https://huggingface.co/nomic-ai/nomic-embed-text-v1.5/blob/1c45008249e960fddc50e443a833bfe147fc1e40/config.json#L8

​

FYI, the "path" in embeddings config actually downloads model to huggingface cache folder just fine. Still the remote code expects it in local folder