CODEWITHSUNDEEP
huggingface-transformerslangchainchromadb
Lucas Azevedo
Lucas Azevedo

Reputation: 2370

Is there a way to use a VectorstoreIndexCreator() with a huggingface cllm model? - Langchain

I'm simply trying to load a document (CSV) but I would like to use a custom llm and not the openAI one.

from langchain.indexes import VectorstoreIndexCreator
from langchain.document_loaders import DataFrameLoader

loader = DataFrameLoader(dataframe, page_content_column="TRANSLATED_COMMENT")
index = VectorstoreIndexCreator().from_loaders([loader])

How can I create a VectorstoreIndexCreator() that uses, for example:

llm = HuggingFaceHub(repo_id='decapoda-research/llama-7b-hf', huggingfacehub_api_token='XXXXX')

Upvotes: 1

Views: 912

Answers (1)

MarcoM
MarcoM

Reputation: 1204

Yes, it's possible. Here's a code snipped to start from:

from langchain.embeddings import HuggingFaceEmbeddings
model_name = "sentence-transformers/paraphrase-multilingual-mpnet-base-v2" # Selecting a sentence embedding model
model_kwargs = {'device': 'cuda'}
encode_kwargs = {'normalize_embeddings': False}
hf_embeddings = HuggingFaceEmbeddings(
    model_name=model_name,
    model_kwargs=model_kwargs,
    encode_kwargs=encode_kwargs
)

index = VectorstoreIndexCreator(embedding=hf_embeddings).from_loaders(loaders)

Upvotes: 1

Related Questions