Upload 4 files

README.md

@@ -1,14 +1,12 @@
 ---
-title: MemoMe
-emoji: 💬
-colorFrom: yellow
-colorTo: purple
+title: RAG PDF Chatbot
+emoji: 📚
+colorFrom: gray
+colorTo: pink
 sdk: gradio
-sdk_version: 5.0.1
+sdk_version: 4.31.0
 app_file: app.py
 pinned: false
-license: mit
-short_description: This is me...
 ---
 
-An example chatbot using [Gradio](https://gradio.app), [`huggingface_hub`](https://huggingface.co/docs/huggingface_hub/v0.22.2/en/index), and the [Hugging Face Inference API](https://huggingface.co/docs/api-inference/index).
+Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

app.py

@@ -1,64 +1,215 @@
+import os
 import gradio as gr
 from huggingface_hub import InferenceClient
 
-"""
-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("deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B")
+# --- LangChain / RAG Imports ---
+from langchain_community.vectorstores import FAISS
+from langchain_community.document_loaders import PyPDFLoader
+from langchain.text_splitter import RecursiveCharacterTextSplitter
+from langchain.chains import ConversationalRetrievalChain
+from langchain.memory import ConversationBufferMemory
+from langchain_community.embeddings import HuggingFaceEmbeddings 
+from langchain_community.llms import HuggingFaceEndpoint
 
+# Global InferenceClient for plain chat (streaming)
+client = InferenceClient("deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B")
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+# ============================================================================
+# PDF Processing & RAG Chain Functions
+# ============================================================================
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+def load_doc(list_file_path):
+    """Load and split PDF documents into chunks."""
+    loaders = [PyPDFLoader(x) for x in list_file_path]
+    pages = []
+    for loader in loaders:
+        pages.extend(loader.load())
+    text_splitter = RecursiveCharacterTextSplitter(
+        chunk_size=1024, 
+        chunk_overlap=64
+    )
+    doc_splits = text_splitter.split_documents(pages)
+    return doc_splits
 
-    messages.append({"role": "user", "content": message})
+def create_db(splits):
+    """Create a vector database from document splits."""
+    # Note: HuggingFaceEmbeddings is deprecated. You may consider using the new package.
+    embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
+    vectordb = FAISS.from_documents(splits, embeddings)
+    return vectordb
 
-    response = ""
+def initialize_database(file_objs):
+    """
+    Process uploaded PDF files, create document splits and a vector database.
+    Expects file objects from gr.Files.
+    """
+    # Each file object's .name attribute holds the file path.
+    list_file_path = [file_obj.name for file_obj in file_objs if file_obj is not None]
+    if not list_file_path:
+        return None, "No files uploaded."
+    doc_splits = load_doc(list_file_path)
+    vector_db = create_db(doc_splits)
+    return vector_db, "Database created!"
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
+def initialize_qa_chain(temperature, max_tokens, top_k, vector_db):
+    """
+    Initialize the retrieval-augmented QA chain using your chat model.
+    An explicit task parameter is passed to avoid the "Task unknown" error.
+    """
+    if vector_db is None:
+        return None, "No vector database available. Please create one first."
+    
+    # Explicitly set the task to "text-generation" to avoid the error.
+    llm = HuggingFaceEndpoint(
+        repo_id="deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B",
+        huggingfacehub_api_token=os.getenv("HF_TOKEN"),  # ensure HF_TOKEN is set
         temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+        max_new_tokens=max_tokens,
+        top_k=top_k,
+        task="text-generation"
+    )
+    
+    memory = ConversationBufferMemory(
+        memory_key="chat_history",
+        output_key='answer',
+        return_messages=True
+    )
+    
+    retriever = vector_db.as_retriever()
+    qa_chain = ConversationalRetrievalChain.from_llm(
+        llm,
+        retriever=retriever,
+        chain_type="stuff", 
+        memory=memory,
+        return_source_documents=True,
+        verbose=False,
+    )
+    return qa_chain, "QA chain initialized. Chatbot is ready!"
+
+def format_chat_history(history):
+    """Format chat history into a list of strings for the QA chain."""
+    formatted = []
+    for user_msg, bot_msg in history:
+        formatted.append(f"User: {user_msg}")
+        formatted.append(f"Assistant: {bot_msg}")
+    return formatted
+
+# ============================================================================
+# Chat Function that switches between plain chat and RAG mode
+# ============================================================================
 
-        response += token
-        yield response
+def chat_respond(message, history, system_message, max_tokens, temperature, top_p, qa_chain):
+    """
+    If a QA chain (i.e. RAG mode) is initialized, use it to generate a response 
+    (including source references). Otherwise, fall back to a plain chat response 
+    using a streaming InferenceClient.
+    """
+    # --- QA Chain (RAG) Mode ---
+    if qa_chain is not None:
+        formatted_history = format_chat_history(history)
+        response = qa_chain.invoke({"question": message, "chat_history": formatted_history})
+        answer = response.get("answer", "")
+        # Optionally include up to 3 source references
+        sources = response.get("source_documents", [])
+        ref_text = ""
+        for i, doc in enumerate(sources[:3]):
+            page = doc.metadata.get("page", "?")
+            ref_text += f"\n\nReference {i+1} (Page {int(page)+1 if page != '?' else '?'}):\n{doc.page_content.strip()}"
+        full_answer = answer + ref_text if ref_text else answer
+        history = history + [(message, full_answer)]
+        return history, qa_chain
 
+    # --- Plain Chat Mode (fallback) ---
+    else:
+        messages = [{"role": "system", "content": system_message}]
+        for user_msg, bot_msg in history:
+            if user_msg:
+                messages.append({"role": "user", "content": user_msg})
+            if bot_msg:
+                messages.append({"role": "assistant", "content": bot_msg})
+        messages.append({"role": "user", "content": message})
+        
+        response = ""
+        result = client.chat_completion(
+            messages,
+            max_tokens=max_tokens,
+            stream=False,
+            temperature=temperature,
+            top_p=top_p,
+        )
+        for token_message in result:
+            token = token_message.choices[0].delta.content
+            response += token
+        
+        history = history + [(message, response)]
+        return history, qa_chain
 
-"""
-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)",
-        ),
-    ],
-)
+# ============================================================================
+# Gradio Interface Layout
+# ============================================================================
 
+with gr.Blocks(theme=gr.themes.Default(primary_hue="sky")) as demo:
+    
+    # States to hold the vector DB and QA chain
+    vector_db_state = gr.State()
+    qa_chain_state = gr.State(None)
+    
+    gr.Markdown("<h1 align='center'>Chat with RAG-enabled PDFs</h1>")
+    gr.Markdown(
+        "Upload PDF files to allow your chatbot to answer questions using information from those documents. "
+        "If no PDFs are uploaded (or the QA chain isn’t initialized), the bot will use plain chat mode."
+    )
+    
+    with gr.Row():
+        with gr.Column(scale=4):
+            gr.Markdown("### Step 1: Document Upload & RAG Setup")
+            pdf_files = gr.Files(file_types=[".pdf"], label="Upload PDF documents")
+            db_status = gr.Textbox(label="Database status", interactive=False)
+            qa_status = gr.Textbox(label="QA Chain status", interactive=False)
+            with gr.Row():
+                create_db_btn = gr.Button("Create Vector DB")
+                init_qa_btn = gr.Button("Initialize QA Chain")
+            top_k_slider = gr.Slider(minimum=1, maximum=10, value=3, step=1, label="Top-k (for RAG)")
+            
+        with gr.Column(scale=6):
+            gr.Markdown("### Step 2: Chat Settings & Conversation")
+            system_message_input = gr.Textbox(value="You are a friendly Chatbot.", label="System Message")
+            max_tokens_slider = gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max New Tokens")
+            temperature_slider = gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature")
+            top_p_slider = gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p")
+            chatbot = gr.Chatbot(label="Chat", elem_id="chatbot", height=400)
+            with gr.Row():
+                user_input = gr.Textbox(placeholder="Enter your message", label="Your Message")
+                send_btn = gr.Button("Send")
+    
+    # -------------------------
+    # Set up button events
+    # -------------------------
+    
+    # Create the vector database from uploaded PDFs.
+    create_db_btn.click(
+        fn=initialize_database,
+        inputs=[pdf_files],
+        outputs=[vector_db_state, db_status]
+    )
+    
+    # Initialize the QA chain (RAG mode) using the vector DB.
+    init_qa_btn.click(
+        fn=initialize_qa_chain,
+        inputs=[temperature_slider, max_tokens_slider, top_k_slider, vector_db_state],
+        outputs=[qa_chain_state, qa_status]
+    )
+    
+    # Chat button: process user input. This function checks if qa_chain is set.
+    send_btn.click(
+        fn=chat_respond,
+        inputs=[user_input, chatbot, system_message_input, max_tokens_slider, temperature_slider, top_p_slider, qa_chain_state],
+        outputs=[chatbot, qa_chain_state]
+    ).then(
+        lambda: "",  # clear the user input box after sending
+        None,
+        user_input
+    )
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.queue().launch()

gitattributes

@@ -0,0 +1,35 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

requirements.txt

@@ -1 +1,9 @@
-huggingface_hub==0.25.2
+torch
+transformers
+sentence-transformers
+langchain
+langchain-community
+tqdm
+accelerate
+pypdf
+faiss-gpu