Upload 4 files

Dockerfile

@@ -0,0 +1,13 @@
+FROM python:3.10.1O
+
+WORKDIR /code
+
+COPY ./requirements.txt /code/requirements.txt
+
+RUN pip install --no-cache-dir --upgrade -r /code/requirements.txt
+
+
+COPY ./rag_retriver.py /code/rag_retriver.py
+
+COPY ./main.py /code/main.py
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

main.py

@@ -0,0 +1,37 @@
+from fastapi import FastAPI, HTTPException
+from pydantic import BaseModel
+from typing import Optional
+
+from rag_retriever import  initialize_llm, initialize_pinecone, create_query_engine, get_response
+
+app = FastAPI()
+
+# Load settings from.env file
+
+
+# Initialize LLM
+initialize_llm()
+
+
+
+# Initialize Pinecone index
+index = initialize_pinecone()
+
+# Create query engine
+query_engine = create_query_engine(index)
+
+class QueryRequest(BaseModel):
+    query: str
+
+@app.post("/query")
+async def query(query_request: QueryRequest):
+    try:
+        response = get_response(query_engine, query_request.query)
+        return {"response": response}
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))
+
+@app.get("/healthcheck")
+async def healthcheck():
+    return {"status": "ok"}
+

rag_retriever.py

@@ -0,0 +1,67 @@
+
+from dotenv import load_dotenv
+import os
+from llama_index.embeddings.huggingface import HuggingFaceEmbedding
+from llama_index.llms.huggingface import HuggingFaceInferenceAPI
+from llama_index.core.settings import Settings
+from llama_index.core import SimpleDirectoryReader
+from llama_index.core.node_parser import SentenceWindowNodeParser
+from pinecone import Pinecone
+from llama_index.core import VectorStoreIndex, StorageContext
+from llama_index.vector_stores.pinecone import PineconeVectorStore
+from llama_index.core.postprocessor import MetadataReplacementPostProcessor
+from llama_index.core.postprocessor import SentenceTransformerRerank
+
+
+def initialize_llm():
+    load_dotenv()
+    HF_TOKEN = os.getenv("HUGGINGFACE_API_KEY")
+    
+    Settings.llm = HuggingFaceInferenceAPI(model_name="mistralai/Mixtral-8x7B-Instruct-v0.1", model_kwargs={"temperature": 0.1}, token=HF_TOKEN)
+    Settings.embed_model = HuggingFaceEmbedding(model_name="BAAI/bge-small-en-v1.5")
+
+
+
+
+def initialize_pinecone():
+    load_dotenv()
+    api_key = os.environ["PINECONE_API_KEY"]
+    index_name = "chatbot"
+    pc = Pinecone(api_key=api_key)
+    pinecone_index = pc.Index(index_name)
+    vector_store = PineconeVectorStore(pinecone_index=pinecone_index)
+    index = VectorStoreIndex.from_vector_store(vector_store=vector_store)
+    return index
+
+def create_query_engine(index):
+    postproc = MetadataReplacementPostProcessor(target_metadata_key="window")
+    rerank = SentenceTransformerRerank(top_n=2, model="BAAI/bge-reranker-base")
+    query_engine = index.as_query_engine(similarity_top_k = 6,vector_store_query_mode="hybrid",alpha=0.5,node_postprocessors = [postproc, rerank],
+)
+    return query_engine
+
+def generate_prompt(context, query):
+    prompt_template = """
+    You are a highly informed representative of KYC SUD CONSULTING. 
+    Your role is to assist users by answering questions about our company with accurate and helpful information. 
+    Use the provided context to guide your responses effectively. 
+    If a question falls outside your knowledge or the context provided, simply state that you don't have the information.
+
+    Given the following context:
+    {context_str}
+
+    Answer the following question in different words while keeping the meaning intact:
+    {query_str}
+
+    Response:
+    """
+    return prompt_template.format(context_str=context, query_str=query)
+
+
+
+# Function to process input sentence and return response
+def get_response(query_engine, sentence):
+    retrieved_context = query_engine.query(sentence)
+    prompt = generate_prompt(retrieved_context, sentence)
+    response = Settings.llm.complete(prompt)  # Use the correct method to generate response
+    return response.text

requirements.txt

@@ -0,0 +1,9 @@
+pinecone-client
+python-dotenv
+llama-index
+llama_index.embeddings.huggingface
+llama-index.llms.huggingface
+llama-index.vector-stores.pinecone
+fastapi
+pydantic
+uvicorn