First_update

app.py

@@ -1,64 +1,91 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
+import faiss
+import numpy as np
+import requests
+import torch
+from transformers import AutoTokenizer, AutoModelForCausalLM
+from sentence_transformers import SentenceTransformer
 
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
+class CustomRetriever:
+    def __init__(self, faiss_index_path: str):
+        """Initializes the retriever by loading the FAISS index and setting up the embedding model."""
+        self.index = faiss.read_index(faiss_index_path)
+        self.embedder = SentenceTransformer('sentence-transformers/all-MiniLM-L6-v2')
+    
+    def retrieve(self, query: str, top_k: int = 5):
+        """Retrieve top-k relevant documents based on the query."""
+        query_embedding = self.embedder.encode([query])
+        distances, indices = self.index.search(np.array(query_embedding).astype('float32'), top_k)
+        return [(index, distance) for index, distance in zip(indices[0], distances[0])]
 
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+class CustomGenerator:
+    def __init__(self):
+        """Initializes the generator by loading the HuggingFace model."""
+        self.tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-Instruct-v0.2")
+        self.model = AutoModelForCausalLM.from_pretrained("mistralai/Mistral-7B-Instruct-v0.2")
+    
+    def generate(self, user_input: str, retrieved_docs: list, max_length: int = 256):
+        """Generate a response using the retrieved documents and the user input."""
+        context = "\n".join([f"Doc {i+1}: {doc}" for i, (doc, _) in enumerate(retrieved_docs)])
+        prompt = f"Context:\n{context}\n\nUser: {user_input}\nBot:"
+        inputs = self.tokenizer(prompt, return_tensors="pt", truncation=True)
+        with torch.no_grad():
+            outputs = self.model.generate(inputs.input_ids, max_length=max_length, pad_token_id=self.tokenizer.eos_token_id)
+        response = self.tokenizer.decode(outputs[0], skip_special_tokens=True)
+        return response.split("Bot:")[-1].strip()
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
 
-    messages.append({"role": "user", "content": message})
+def rag_chatbot(user_input):
+    """The main RAG chatbot function to retrieve documents and generate a response."""
+    top_k = 5  # Number of documents to retrieve
+    retrieved_doc_ids = retriever.retrieve(user_input, top_k)
+    retrieved_docs = [(f"Dummy content for doc {doc_id}", distance) for doc_id, distance in retrieved_doc_ids]
+    response = generator.generate(user_input, retrieved_docs)
+    return response
 
-    response = ""
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+FAISS_INDEX_PATH = "path/to/your/faiss_index"
+retriever = CustomRetriever(faiss_index_path=FAISS_INDEX_PATH)
+generator = CustomGenerator()
 
-        response += token
-        yield response
+# Gradio UI
+app = gr.Blocks()
 
+with app:
+    gr.Markdown(
+        """
+        # Banking Regulations Compliance ChatBOT
+        Ask questions and get responses generated by a state-of-the-art AI model!
+        """
+    )
 
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
-    ],
-)
+    chatbot = gr.Chatbot(label="Chat with the Bot")
 
+    question_box = gr.Textbox(
+        label="Your Message",
+        placeholder="Type your message here...",
+        lines=1,
+    )
+    submit_button = gr.Button("Send")
 
-if __name__ == "__main__":
-    demo.launch()
+    gr.Examples(
+        examples=[
+            "What is Compliance?",
+            "Can you summarize the RBI Guidelines?",
+            "Can you summarize the RBI Guidelines related to gold loans?",
+        ],
+        inputs=question_box,
+    )
+
+    submit_button.click(ask_model, inputs=[chatbot, question_box], outputs=chatbot)
+
+    footer_md = """
+    ---
+    © 2024 by [Shashi Kiran, Karthik K, Venkara V V, Navin Kumar N, Jyoti Bavne]. All rights reserved.
+    This app was developed by **6505 Project Team** as part of Final Project.
+    For inquiries, please contact: [schandrappa@student.fairfield.edu](mailto:schandrappa@student.fairfield.edu)
+    """
+    gr.Markdown(footer_md)
+
+app.launch()