app.py

import gradio as gr
from llama_index import SimpleDirectoryReader, ServiceContext
from llama_index.llms import HuggingFaceLLM
from langchain.embeddings.huggingface import HuggingFaceEmbeddings
from llama_index.embeddings import LangchainEmbedding
import torch
import chromadb
from chromadb.utils import embedding_functions

# Configure Llama2 Model
llm = HuggingFaceLLM(
    context_window=4096,
    max_new_tokens=256,
    generate_kwargs={"temperature": 0.0, "do_sample": False},
    tokenizer_name="meta-llama/Llama-2-7b-chat-hf",
    model_name="meta-llama/Llama-2-7b-chat-hf",
    device_map="auto",
    model_kwargs={"torch_dtype": torch.float16 , "load_in_8bit": True}
)

# Set Up Embedding Model
embed_model = LangchainEmbedding(
    HuggingFaceEmbeddings(model_name="sentence-transformers/all-mpnet-base-v2")
)

# Create Query Function
def query_chatbot(query):
    service_context = ServiceContext.from_defaults(
        chunk_size=1024,
        llm=llm,
        embed_model=embed_model
    )
    
    # Initialize ChromaDB
    client = chromadb.Client()
    collection = client.create_collection(
        "documents", 
        embedding_function=embedding_functions.HuggingFaceEmbeddingFunction(model_name="sentence-transformers/all-mpnet-base-v2")
    )
    
    # Load documents (you can customize this part)
    documents = SimpleDirectoryReader("/path/to/your/data").load_data()
    
    # Use GPTSimpleIndex as an alternative to VectorStoreIndex
    from llama_index.indices import GPTSimpleIndex
    index = GPTSimpleIndex.from_documents(documents, service_context=service_context, vector_store=collection)
    
    # Create Query Engine
    query_engine = index.as_query_engine()
    response = query_engine.query(query)
    
    return response

# Create Gradio Interface
iface = gr.Interface(
    fn=query_chatbot, 
    inputs="text", 
    outputs="text", 
    title="Blog Assistant Chatbot"
)

iface.launch()