In this blog, we’ll build a real-time text summarizer using DeepSeek-R1, LangChain, and Ollama, all integrated into a Streamlit app. The goal is to stream responses in real-time so users can see the summary being generated as it happens.

Why Use Ollama and DeepSeek-R1?

Ollama provides local access to powerful AI models like DeepSeek-R1, eliminating the need for cloud-based APIs. DeepSeek-R1 is a strong open-weight LLM, great for tasks like text summarization.

Setting Up the Project

1. Install Dependencies

pip install streamlit langchain-ollama

2. Start Ollama and Download the Model

ollama serve
ollama pull deepseek-r1:1.5b

3. Create the Streamlit App (app.py)

The following code initializes a LangChain Ollama model, sets up a streaming response, and displays the summary in real-time in the Streamlit UI.

pythonCopyEditimport streamlit as st
from langchain_ollama.chat_models import ChatOllama
from langchain.schema import HumanMessage

st.set_page_config(page_title="Text Summarizer", layout="centered")
st.title("📝 AI Text Summarizer (Streaming)")
st.write("Using DeepSeek-R1 with LangChain & Ollama for real-time summarization")

user_input = st.text_area("Enter the text you want to summarize:", height=200)

llm = ChatOllama(
    model="deepseek-r1:1.5b",
    temperature=0.7,
    base_url="http://localhost:11434",
    streaming=True,
)

def generate_summary(input_text):
    messages = [HumanMessage(content=f"Summarize the following text:\n\n{input_text}\n")]
    for chunk in llm.stream(messages):
        yield chunk.content  

if st.button("Summarize"):
    if user_input.strip():
        with st.spinner("Generating summary..."):
            summary_placeholder = st.empty()
            summary_text = ""

            for chunk in generate_summary(user_input):
                summary_text += chunk
                summary_placeholder.markdown(summary_text)
    else:
        st.warning("Please enter some text to summarize.")

How It Works

1. User inputs text for summarization.
2. The LangChain Ollama model (DeepSeek-R1) processes the input.
3. The summary streams in real-time as it is being generated.

Running the App

Start the Streamlit app with:

streamlit run app.py

Now, you’ll see the summary update dynamically instead of waiting for the full response! 🚀

Conclusion

With Ollama, DeepSeek-R1, and LangChain, we created a real-time AI summarization app that runs locally. This is a great way to build efficient, interactive AI applications without relying on external APIs.

Would you like to see more AI-powered Streamlit apps? Let me know in the comments! 😊