resolve dependency issues

.env

@@ -0,0 +1,3 @@
+OPENAI_API_KEY = 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
+QDRANT_URL = https://63f1dc5d-951d-4539-be73-94d4f725452f.us-east4-0.gcp.cloud.qdrant.io:6333
+QDRANT_API_KEY = x_KgU_SErG-GhKnhGaoi94S4Lrna83s-rwFdcqOyUJbHGVAtgmPflQ

.gitignore

@@ -0,0 +1,2 @@
+*.ipynb
+virt/**/*

app.py

@@ -0,0 +1,81 @@
+import gradio as gr
+from langchain.vectorstores import Qdrant
+import qdrant_client
+from langchain.schema.runnable import RunnablePassthrough
+from langchain.schema.output_parser import StrOutputParser
+from langchain.prompts import ChatPromptTemplate
+from langchain.chat_models import ChatOpenAI
+from langchain.embeddings import OpenAIEmbeddings
+import dotenv
+import os
+from utils import template
+import time
+
+# Load environment variables and validate
+dotenv.load_dotenv()
+QDRANT_URL = os.getenv("QDRANT_URL")
+QDRANT_API_KEY = os.getenv("QDRANT_API_KEY")
+if not QDRANT_URL or not QDRANT_API_KEY:
+    raise ValueError("QDRANT_URL and QDRANT_API_KEY must be set in the environment")
+
+# Initialize the vector store
+def initiliaze_vector_store():
+    """
+    Initialize and return the vector store.
+    Only run this on launch.
+    """
+    embeddings = OpenAIEmbeddings()
+    client = qdrant_client.QdrantClient(url=QDRANT_URL, api_key=QDRANT_API_KEY)
+    vectorstore = Qdrant(client=client, collection_name="doc_datategy", embeddings=embeddings)
+    return vectorstore
+
+# Initialize the retriever
+def initiliaze_retriever(vectorstore):
+    """
+    Initialize and return the retriever using the given vectorstore.
+    """
+    return vectorstore.as_retriever()
+
+# Initialize the chatbot
+def initiliaze_chatbot(template, model_name="gpt-3.5-turbo-1106", temperature=0):
+    """
+    Initialize and return the chatbot components: prompt and language model.
+    """
+    prompt = ChatPromptTemplate.from_template(template)
+    llm = ChatOpenAI(model_name=model_name, temperature=temperature)
+    return prompt, llm
+
+# Initialize the RAG chain
+def initiliaze_RAG(retriever, prompt, llm):
+    """
+    Initialize and return the RAG chain.
+    """
+    context_function = {"context": retriever, "question": RunnablePassthrough()}
+    rag_chain = context_function | prompt | llm | StrOutputParser()
+    return rag_chain
+
+# Launch Gradio app
+vectorstore = initiliaze_vector_store()
+retriever = initiliaze_retriever(vectorstore)
+
+with gr.Blocks() as demo:
+    chatbot = gr.Chatbot(label="PapAI custom chatbot")
+    msg = gr.Textbox(label="Prompt", value='PapAI?', interactive=True)
+    clear = gr.Button("Clear")
+    template_user = gr.Textbox(label="Template", value=template, interactive=True)
+
+    def change_template(template_user_str):
+        prompt, llm = initiliaze_chatbot(template_user_str)
+        return initiliaze_RAG(retriever, prompt, llm)
+
+    def RAG_answer(query, chat_history, template_user_str):
+        rag_chain = change_template(template_user_str)
+        answer = rag_chain.invoke(query)
+        chat_history.append((query, answer))
+        time.sleep(1.3)  # Consider optimizing or dynamic handling
+        return "", chat_history
+
+    msg.submit(RAG_answer, [msg, chatbot, template_user], [msg, chatbot])
+
+demo.queue()
+demo.launch(share=False, debug=True)

utils.py

@@ -0,0 +1,19 @@
+template = """You are an assistant for question-answering tasks. 
+Use the following pieces of retrieved context to answer the question. 
+If you don't know the answer, just say that you don't know. 
+Use three sentences maximum and keep the answer concise.
+Question: {question} 
+Context: {context} 
+Answer:
+"""
+
+template = """You are an assistant specialized in answering (specify type, e.g., data analysis/statistics/machine learning) questions for Data Science projects. Follow these steps to answer:
+1. Verify if the provided context is relevant to the question. If not, request more specific information or additional context.
+2. Use the following pieces of retrieved context to answer the question. Incorporate data-driven insights or references when possible.
+3. If the answer requires external data science tools or datasets, mention this in your response.
+4. If you don't know the answer or if the information is insufficient to formulate an answer, clearly state that.
+5. Keep your answer concise, using a maximum of three sentences.
+Question: {question} 
+Context: {context} 
+Answer:
+"""