Chatbot (Ollama and Langchain)

Overview

Built using the Langchain framework and Ollama, this chatbot helps me explore the topic of AI and large language models (LLMs).

Key Features

• Context memory: The chatbot remembers past conversations, so it gives more relevant responses.
• Customization: If you use DeepSeek R1, you can choose to show or hide its thinking using the /toggle_thoughts command.

Installation

Prerequisites:

• Docker Desktop for Windows
• Windows Subsystem for Linux (WSL)
• Internet connection
• At least 15GB of storage

Follow this procedure to install.

1. Clone the repository.

   git clone https://github.com/boysbytes/langchain-chatbot.git
   cd langchain-chatbot

2. Create the Ollama data directory.

   mkdir data
   mkdir ./data/ollama

3. Build the Docker image.

   docker-compose build

4. Start the services.

   docker-compose up -d

5. Pull a model.

   docker exec -it ollama ollama run deepseek-r1:1.5b

Usage

1. Access the chatbot: http://localhost:3000

2. By default, its thinking process is hidden. You can see its thinking process by sending this command: /toggle_thoughts

3. When you're done, stop and clean-up.

   docker-compose down

Configuration

Change the model

1. Start the Docker services.

2. Replace the model.

   # List the models
   docker exec -it ollama ollama list
   
   # Delete a model
   docker exec -it ollama ollama rm <model_name>
   
   # Pull another model
   docker exec -it ollama ollama run <model_name> 

3. Update the model name in chatbot.py

   class Chatbot:
       def __init__(self, model_name="<model_name>", temperature=0.6):
           # Initialize memory to store conversation history
           self.memory = ConversationBufferMemory(return_messages=True)

   For example, if you use deepseek-r1:1.5b, then update as follows:

   class Chatbot:
       def __init__(self, model_name="deepseek-r1:1.5b", temperature=0.6):
           # Initialize memory to store conversation history
           self.memory = ConversationBufferMemory(return_messages=True)

4. Rebuild and run the containers.

   docker-compose up --build

   Troubleshooting

• Ensure Docker Desktop is running.

• Check container logs: If something doesn't work, monitor the logs for errors:

  docker-compose logs -f

• Verify the model was pulled successfully.

  docker exec ollama-chatbot ollama list

• Restart services: If issues persist, restart the containers:

  docker-compose down
  docker-compose up -d