🤖 DeepSeek-R1 LangChain-Ollama Chat | Gradio UI

Click above to open the fully implemented notebook in Colab ☝️

Interactive environment to explore DeepSeek-R1 models on Google Colab, featuring Gradio UI and LangChain pipelines, using Ollama as the model runtime.

Available models:

• deepseek-r1:1.5b (default - distilled from Qwen-2.5-Math-1.5B)
• deepseek-r1:7b (distilled from Qwen-2.5-Math-7B)
• deepseek-r1:8b (distilled from Llama-3.1-8B)
• deepseek-r1:14b (distilled from Qwen-2.5-14B)

All models are compatible with Google Colab's free T4 GPU (16GB).

✨ Features

• Multiple DeepSeek-R1 models (1.5B to 14B variants for free T4 GPU)
• Chat interface with Gradio
• LangChain conversation pipeline
• XML parsing for reasoning separation
• Automated Ollama setup

🚀 Usage

1. Open notebook in Colab
2. Set T4 GPU runtime
3. Run cells in order

You can also find the Jupyter notebook in this repo: deepseek_r1_gradio_env.ipynb

📋 Requirements

• Google Colab (T4 GPU)
• GPU RAM: 16GB

🔗 Resources

• DeepSeek R1 Repository
• DeepSeek R1 on Ollama
• Gradio
• LangChain
• Ollama
• LLM-XML-Parser

---

Created by Edoardo Avenia