app.py

from langchain.text_splitter import CharacterTextSplitter
from langchain.embeddings import HuggingFaceEmbeddings
from langchain.vectorstores import Chroma
from langchain import HuggingFacePipeline
from langchain.chains import RetrievalQA
from transformers import AutoTokenizer
import pickle
import os
from langchain.document_loaders import BSHTMLLoader, DirectoryLoader

bshtml_dir_loader = DirectoryLoader('./data/', loader_cls = BSHTMLLoader)

data = bshtml_dir_loader.load()

with open("shakespeare.pkl", "wb") as fp:
  pickle.dump(data, fp)

with open('shakespeare.pkl', 'rb') as fp:
    data = pickle.load(fp)

bloomz_tokenizer = AutoTokenizer.from_pretrained('bigscience/bloomz-1b7')

text_splitter = CharacterTextSplitter.from_huggingface_tokenizer(bloomz_tokenizer, chunk_size=100, chunk_overlap=0, separator='\n')

documents = text_splitter.split_documents(data)

embeddings = HuggingFaceEmbeddings()

persist_directory = "vector_db"

vectordb = Chroma.from_documents(documents=documents, embedding=embeddings, persist_directory=persist_directory)

vectordb.persist()
vectordb = None

vectordb_persist = Chroma(persist_directory=persist_directory, embedding_function=embeddings)

llm = HuggingFacePipeline.from_model_id(
    model_id="bigscience/bloomz-1b7",
    task="text-generation",
    model_kwargs={"temperature" : 0, "max_length" : 500})

doc_retriever = vectordb_persist.as_retriever()

shakespeare_qa = RetrievalQA.from_chain_type(llm=llm, chain_type="stuff", retriever=doc_retriever)

def make_inference(query):
    inference = shakespeare_qa.run(query)
    return inference

if __name__ == "__main__":
    # make a gradio interface
    import gradio as gr

    gr.Interface(
        make_inference,
        gr.inputs.Textbox(lines=2, label="Query"),
        gr.outputs.Textbox(label="Response"),
        title="Ask_Shakespeare",
        description="️building_w_llms_qa_Shakespeare allows you to inquire about the Shakespeare's plays.",
    ).launch()