gpt4all with knowledge

main.py

from langchain.llms import GPT4All
from langchain.vectorstores import Chroma
from langchain.indexes import VectorstoreIndexCreator
from langchain.indexes.vectorstore import VectorStoreIndexWrapper
from langchain.document_loaders import DirectoryLoader, TextLoader, UnstructuredMarkdownLoader
from langchain.chains import RetrievalQA
import sys
import os
from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
from langchain.embeddings import HuggingFaceEmbeddings, SentenceTransformerEmbeddings

# https://gist.github.com/scriptsandthings/75c38c54e05dd20d65fd83a9bd522406


# GPT4ALL_MODEL_PATH = r'C:\Users\spenc\.models\GPT4All-13B-snoozy.ggmlv3.q4_0.bin' # works
# embeddings = LlamaCppEmbeddings(model_path=GPT4ALL_MODEL_PATH)

GPT4ALL_MODEL_PATH = r'C:\Users\spenc\.models\GPT4All-13B-snoozy.ggmlv3.q4_0.bin' # works
embeddings = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")

def main() -> None:
    print('GPT Deki')

    # Enable to save to disk & reuse the model (for repeated queries on the same data)
    PERSIST = True

    query = "What is the name of my dog?"

    if PERSIST and os.path.exists("persist"):
        print("Reusing index...\n")
        vectorstore = Chroma(persist_directory="persist",embedding_function=embeddings)
        index = VectorStoreIndexWrapper(vectorstore=vectorstore)
    else:
        loader = TextLoader("data.txt") # Use this line if you only need data.txt
        # loader = DirectoryLoader("data/")
        if PERSIST:
            index = VectorstoreIndexCreator(embedding=embeddings,vectorstore_kwargs={"persist_directory":"persist"}).from_loaders([loader])
        else:
            index = VectorstoreIndexCreator(embedding=embeddings).from_loaders([loader])

    model = GPT4All(model=GPT4ALL_MODEL_PATH)
    chain = RetrievalQA.from_chain_type(llm=model,retriever=index.vectorstore.as_retriever(search_kwargs={"k": 1}))
    print(chain.run(query))

requirements.txt

langchain
chromadb
tiktoken
gpt4all
llama-cpp-python
sentence-transformers