Update app.py

app.py

@@ -1,204 +1,61 @@
-import streamlit as st 
-import os
-from langchain_huggingface import HuggingFaceEndpoint, HuggingFacePipeline, ChatHuggingFace
-from langchain_core.messages import HumanMessage,SystemMessage,AIMessage
-
-hf_token = os.getenv("HF_TOKEN")
-
-
-# Set the default page in session state
-if "page" not in st.session_state:
-    st.session_state.page = "home"
-
-# Function to switch pages
-def switch_page(page_name):
-    st.session_state.page = page_name
-
-# Home page with buttons for different domains
-if st.session_state.page == "home":
-    st.title("🤖 Innomatics ChatGenius Hub")
-    st.markdown("Choose a domain to chat with an expert model:")
-
-    col1, col2, col3 = st.columns(3)
-    with col1:
-        if st.button("Python 🐍"):
-            switch_page("python")
-        if st.button("Statistics 📈"):
-            switch_page("statistics")
-        
-        
-
-    with col2:
-        if st.button("SQL 🛢️"):
-            switch_page("sql")
-        if st.button("Machine Learning 🤖"):
-            switch_page("ml")
-        
-
-    with col3:
-        if st.button("Power BI 📊"):
-            switch_page("powerbi")
-        if st.button("Deep Learning 🧠"):
-            switch_page("deeplearning")
-    with col2:
-        if st.button("GenAI🔮🤖"):
-            switch_page("genai")
-            
-        
-
-# Example domain-specific chatbot page
-elif st.session_state.page == "python":
-    st.title("Python Chatbot 🐍")
-#     hf_token = os.getenv("HUGGINGFACEHUB_API_TOKEN") or os.getenv("HF_TOKEN")
-#     if not hf_token:
-#         st.error("Please add your Hugging Face API token to Secrets (HUGGINGFACEHUB_API_TOKEN or HF_TOKEN).")
-#         st.stop()
-
-# # Setup the LangChain HuggingFaceEndpoint and ChatHuggingFace LLM
-#     deep_seek_model = HuggingFaceEndpoint(
-#     repo_id="deepseek-ai/DeepSeek-R1",
-#     # provider = 'nebius'
-#     temperature=0.7,
-#     max_new_tokens=100,
-#     task="conversational",
-#     huggingfacehub_api_token=hf_token,
-#     )
-
-#     deepseek = ChatHuggingFace(
-#     llm=deep_seek_model,
-#     repo_id="deepseek-ai/DeepSeek-R1",
-#     # provider="nebius",
-#     temperature=0.7,
-#     max_new_tokens=100,
-#     task="conversational"
-#     )
-
-
-    gemma_model = HuggingFaceEndpoint(
-    repo_id="google/gemma-3-27b-it",
-    temperature=0.7,
-    max_new_tokens=512,
-    task="conversational",
-    huggingfacehub_api_token=hf_token,
+import streamlit as st
+from langchain_community.document_loaders import YoutubeLoader
+from transformers import AutoTokenizer, AutoModelForCausalLM
+import torch
+
+st.set_page_config(page_title="DeepSeek YouTube Summarizer", layout="centered")
+st.title("📺 YouTube Video Summarizer with DeepSeek")
+
+# Input YouTube URL
+url = st.text_input("Enter YouTube Video URL:")
+
+# Load DeepSeek model and tokenizer
+@st.cache_resource
+def load_model():
+    model_id = "deepseek-ai/deepseek-llm-7b-instruct"
+    tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
+    model = AutoModelForCausalLM.from_pretrained(
+        model_id, device_map="auto", torch_dtype=torch.float16, trust_remote_code=True
     )
-
-    chat_gemma = ChatHuggingFace(
-    llm=gemma_model,
-    repo_id="google/gemma-3-27b-it",
-    temperature=0.7,
-    max_new_tokens=512,
-    task="conversational",
+    return tokenizer, model
+
+tokenizer, model = load_model()
+
+def summarize_with_deepseek(text):
+    prompt = f"""<|system|>\nYou are a helpful assistant.\n<|user|>\nSummarize the following text:\n{text}\n<|assistant|>"""
+    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+    outputs = model.generate(
+        **inputs,
+        max_new_tokens=300,
+        do_sample=False,
+        temperature=0.7,
+        top_k=50,
+        top_p=0.95,
     )
-# Initialize session state for chat history
-    if "messages" not in st.session_state:
-        st.session_state.messages = [
-        SystemMessage(content="Answer like a 10 year experinced Python developer")
-    ]
-
-    def generate_response(user_input):
-    # Append user message
-        st.session_state.messages.append(HumanMessage(content=user_input))
-    # Invoke the model
-        response = chat_gemma.invoke(st.session_state.messages)
-    # Append AI response
-        st.session_state.messages.append(AIMessage(content=response))
-        return response
-
-# User input
-    user_input = st.text_input("Ask a question about Python:")
-
-    if user_input:
-        with st.spinner("Getting answer..."):
-            answer = generate_response(user_input)
-        st.markdown(f"**Answer:** {answer}")
-
-# Display chat history
-    if st.session_state.messages:
-        for msg in st.session_state.messages[1:]:  # skip initial SystemMessage
-            if isinstance(msg, HumanMessage):
-                st.markdown(f"**You:** {msg.content}")
-            elif isinstance(msg, AIMessage):
-                st.markdown(f"**Bot:** {msg.content}")
-    st.button("⬅️ Back to Home", on_click=lambda: switch_page("home"))
-    # Here you can load your Python LLM and chat interface
-
-elif st.session_state.page == "sql":
-    st.title("SQL Chatbot 🛢️")
-    if not hf_token:
-        st.error("Please add your Hugging Face API token as an environment variable.")
-        st.stop()
-
-# Initialize the LLaMA model from HuggingFace (via Nebius provider)
-    llama_model = HuggingFaceEndpoint(
-    repo_id="meta-llama/Llama-3.1-8B-Instruct",
-    temperature=0.7,
-    max_new_tokens=512,
-    task="conversational",
-    huggingfacehub_api_token=hf_token,
-    )
-
-    llama = ChatHuggingFace(
-    llm=llama_model,
-    repo_id="meta-llama/Llama-3.1-8B-Instruct",
-    # provider="nebius",
-    temperature=0.7,
-    max_new_tokens=512,
-    task="conversational"
-    )
-
-# Streamlit A
-
-    st.markdown("Ask anything related to SQL interviews!")
-
-# Initialize chat history
-    if "messages" not in st.session_state:
-        st.session_state.messages = [SystemMessage(content="Answer clearly like a technical 10 year experienced person in SQL .")]
-
-# User input
-    user_input = st.text_input("💡 Ask your SQL interview question:", placeholder="e.g., give me 10 SQL interview questions with answers")
-
-    def generate_response(user_input):
-        st.session_state.messages.append(HumanMessage(content=user_input))
-        response = llama.invoke(st.session_state.messages)
-        st.session_state.messages.append(AIMessage(content=response))
-        return response
-
-# Display response
-    if user_input:
-        with st.spinner("Thinking..."):
-            answer = generate_response(user_input)
-            st.markdown(f"**Answer:** {answer}")
-
-# Show chat history
-    st.markdown("### 📜 Chat History")
-    for msg in st.session_state.messages[1:]:  # Skip SystemMessage
-        if isinstance(msg, HumanMessage):
-            st.markdown(f"**You:** {msg.content}")
-        elif isinstance(msg, AIMessage):
-            st.markdown(f"**Bot:** {msg.content}")
-        st.button("⬅️ Back to Home", on_click=lambda: switch_page("home"))
-        # Load SQL chatbot here
-
-elif st.session_state.page == "powerbi":
-    st.title("Power BI Chatbot 📊")
-    st.button("⬅️ Back to Home", on_click=lambda: switch_page("home"))
-
-elif st.session_state.page == "ml":
-    st.title("Machine Learning Chatbot 🤖")
-    st.button("⬅️ Back to Home", on_click=lambda: switch_page("home"))
-
-elif st.session_state.page == "deeplearning":
-    st.title("Deep Learning Chatbot 🧠")
-    st.button("⬅️ Back to Home", on_click=lambda: switch_page("home"))
+    summary = tokenizer.decode(outputs[0], skip_special_tokens=True)
+    return summary.split("<|assistant|>")[-1].strip()
 
-elif st.session_state.page == "statistics":
-    st.title("Statistics Chatbot 📈")
-    st.button("⬅️ Back to Home", on_click=lambda: switch_page("home"))
+if st.button("Extract and Summarize"):
+    if url:
+        try:
+            loader = YoutubeLoader.from_youtube_url(url)
+            data = loader.load()
+            transcript = data[0].page_content if data else "No transcript found."
 
-elif st.session_state.page == "genai":
-    st.title("GenAI Chatbot 📈")
-    st.button("⬅️ Back to Home", on_click=lambda: switch_page("home"))
+            st.subheader("📖 Extracted Transcript")
+            st.text_area("Transcript:", transcript, height=300)
 
+            # Truncate for prompt length safety
+            if len(transcript) > 1500:
+                transcript = transcript[:1500]
 
+            with st.spinner("Summarizing using DeepSeek..."):
+                summary = summarize_with_deepseek(transcript)
 
+            st.subheader("🧠 Summary")
+            st.success(summary)
 
+        except Exception as e:
+            st.error(f"Error: {str(e)}")
+    else:
+        st.warning("Please enter a valid YouTube URL.")