Find answers from the community

Updated 6 months ago

Custom models

At a glance

hey all, i had a notebook running with flan and hugginf_face embeddings as:

Plain Text

hfemb = HuggingFaceEmbeddings()
embed_model = LangchainEmbedding(hfemb)

and then

Plain Text

index = GPTSimpleVectorIndex.from_documents(documents, embed_model=embed_model, llm_predictor=llm_predictor)

but I believe that usage of embed_model is deprecated? or else what am I missing?

5 comments

LLogan M

You'll want to set llm_predictor and embed_model inside the service context object, and then pass the service context object into the index constructor 👍

LLogan M

Partial example here with the llm predictor

LLogan M

https://gpt-index.readthedocs.io/en/latest/how_to/customization/custom_llms.html#example-using-a-custom-llm-model

LLogan M

(Also note that flans max input size is 512, so adjust the prompt helper accordingly)

ttinoargentino

Oh nice thanks I'll try that

Add a reply