DeepSeek R1 Setup on Google Colab using Ollama Server

This guide will walk you through the steps to set up and run the DeepSeek R1 model on Google Colab using the Ollama server. Additionally, you’ll learn how to create a Gradio UI to interact with the model in a user-friendly way.

---

Prerequisites

• A Google account (to access Google Drive and Google Colab).
• Basic familiarity with Google Colab and terminal commands.

---

Step 1: Create a New Folder in Google Drive

1. Go to Google Drive and log in.
2. Click on the + New button and select Folder.
3. Name the folder (e.g., DeepSeek-R1-Project).

---

Step 2: Create a Google Colab Notebook

1. Open the folder you just created.
2. Right-click inside the folder, select More, and then choose Google Colaboratory.
3. A new Colab notebook will be created. Name it (e.g., DeepSeek-R1-Notebook).

---

Important: Enable GPU/TPU for Better Performance

1. To improve the model's performance, you can enable GPU or TPU:
   • Go to Runtime > Change runtime type.
   • Under Hardware accelerator, select GPU or TPU.
   • Click Save and restart the notebook.

---

Step 3: Install Required Libraries

1. In the Colab notebook, paste the following commands to install the necessary libraries:

   !pip install langchain
   !pip install langchain-core
   !pip install langchain-community
   !pip install colab-xterm

2. Run the cells to install the libraries.

---

Step 4: Load Colab-Xterm and Open a Terminal

1. Paste the following commands to load colab-xterm and open a terminal:

   %load_ext colabxterm
   %xterm

2. A terminal will open in a new tab within Colab.

---

Step 5: Install Ollama

1. In the terminal, run the following command to install Ollama:

   curl -fsSL https://ollama.com/install.sh | sh

2. Wait for the installation to complete.

---

Step 6: Start Ollama Server and Run DeepSeek R1

1. In the terminal, start the Ollama server and load the DeepSeek R1 model by running the following commands:

   ollama serve &

   ollama run deepseek-r1:7b

2. The DeepSeek R1 model will be loaded and ready to use.

---

Step 7: Add a Gradio UI for Interaction

To make it easier to interact with the DeepSeek R1 model, you can add a Gradio UI. Follow these steps:

Step 7.1: Install Gradio

1. Install the Gradio library by running the following command in a Colab cell:

   !pip install gradio

Step 7.2: Create a Function to Interact with the Model

1. Define a function that sends user prompts to the DeepSeek R1 model and returns the response. Paste the following code into a Colab cell:

   import subprocess
   
   def query_deepseek_r1(prompt):
       command = f'ollama run deepseek-r1:7b "{prompt}"'
       result = subprocess.run(command, shell=True, capture_output=True, text=True)
       
       return result.stdout

Step 7.3: Build the Gradio Interface

1. Create a Gradio interface using the function above. Paste the following code into a Colab cell:

   import gradio as gr
   
   def gradio_interface(prompt):
       response = query_deepseek_r1(prompt)
       return response
   
   iface = gr.Interface(
       fn=gradio_interface,  
       inputs="text",        
       outputs="text",       
       title="DeepSeek R1 Chatbot",  
       description="Ask anything to the DeepSeek R1 model!"
   )
   
   iface.launch(share=True)

2. Run the cell. Gradio will generate a public link to your UI.

Step 7.4: Interact with the Model via Gradio

1. Open the Gradio UI link in your browser.
2. Type a prompt (e.g., "What is the capital of France?") in the input box.
3. Click Submit.
4. The DeepSeek R1 model will process the prompt and display the response in the output box.

---

---

Step 8: Save Your Work

1. To save your Colab notebook, go to File > Save a copy in Drive.
2. Choose the folder you created earlier (e.g., DeepSeek-R1-Project).

---

Troubleshooting

• If the Ollama server fails to start, ensure that the installation was successful and try running the commands again.
• If you encounter any issues with the model, check your internet connection and ensure that you have sufficient resources (e.g., GPU/TPU enabled).
• If the Gradio UI does not load, ensure that the share=True parameter is set and that your Colab runtime is active.

---

Conclusion

You have successfully set up and run the DeepSeek R1 model on Google Colab using the Ollama server and added a Gradio UI for easy interaction. Enjoy experimenting with the model!

---

If you have any questions or run into issues, let me know! 😊