Langchain python

Vectara Platform Q&A
1

Team,
I am getting error while using the following langchain code

"
from langchain.vectorstores import Vectara
loader = TextLoader(“myfile.txt”)
documents = loader.load()
vectara = Vectara.from_documents(documents)
qa = RetrievalQA.from_llm(llm=OpenAI(), retriever=vectara.as_retriever())
"
Error:

vectara = Vectara.from_documents(documents)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
TypeError: VectorStore.from_documents() missing 1 required positional argument: ‘embedding’

According to langchain document it says Vectara provides its own embeddings.

How can I fix this?

Thanks
Ramarao

2

I think you have to provide a blank embedding to satisfy the current API requirements.

3

Thanks. I had to use openai embeddings as I am using retrievalqachain.

Regards
Ramarao