Ambot Link

A scalable and robust Retrieval-Augmented Generation (RAG) system designed for the AMBER (https://ambermd.org/) AI chatbot.

Description

Ambot is designed to assist users in resolving technical issues related to AMBER installation, simulation setup, and troubleshooting efficiently. Instead of spending hours searching through online forums, documentation, or the AMBER mailing list, users can directly ask their questions to Ambot. It leverages the AMBER manual and Q&A from the AMBER mailing list (http://archive.ambermd.org/) to save time and providing reliable solutions.

Whether you're facing errors during installation, struggling with simulation setup, or encountering runtime issues, Ambot is here to help. It serves as a one-stop solution for AMBER-related queries, making it easier for researchers and developers to focus on their work rather than troubleshooting.

Getting Started

Dependencies

• Python: 3.11
• OS: Ubuntu 22.04.4 LTS
• pip: 23.2.1
• Nvidia Driver Version: 555.42
• CUDA Version: 12.5, 12.6, 12.8

LLM Providers

• deepseek-r1:14b

Text embed provider

• nomic-embed-text:latest

Installation Instructions (Ubuntu)

Follow these steps to install and set up Ambot:

• Open your terminal and run the command to clone the repository.

git clone https://github.com/NgFEP/Ambot.git

• Change the directory

cd Ambot

python3 -m venv Ambot_env

source Ambot_env/bin/activate

• Ensure you have Python 3.11+ installed, then run:

pip install -r requirements.txt

• Download Ollama (https://ollama.com/download)

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

These commands are used to manage the Ollama service (likely a machine learning model server) on a Linux system using systemctl, the systemd service manager.(https://github.com/ollama/ollama/blob/main/docs/linux.md)

sudo systemctl daemon-reload
sudo systemctl enable ollama
sudo systemctl start ollama
sudo systemctl status ollama

To pull the deepseek-r1:14b and nomic-embed-text:latest models using Ollama,

ollama pull deepseek-r1:14b

ollama pull nomic-embed-text:latest

• Create a folder to store the embeddings.

mkdir -p embeddings

Usage

Step 1

• To generate embeddings and create Ambot:

python embedding_generation.py

Step 2

• To query and test Ambot:

python query_retrieval.py

You'll be prompted to enter a query:

Enter your query (or type 'exit' to quit):

For an example:

Changing the Large Language Model (LLM)

To switch from the "deepseek-r1:14b" LLM to "Llama 3.2", you need to modify the LLM_MODEL setting in embedding_generation.py and query_retrieval.py.

LLM_MODEL = "deepseek-r1:14b"

to

LLM_MODEL = "Llama 3.2"

Installation Instructions (Windows with Anaconda)

Follow these steps to install and set up Ambot on Windows using Anaconda:

1. Install Anaconda
   Download and install Anaconda from official website. if you haven't already.

2. Install CUDA Toolkit and Add to System PATH

   • Download CUDA Toolkit from NVIDIA.

   • Choose the version 12.8, 12.6 or 12.5 preferably.

   • During installation, select Custom Options and ensure Visual Studio Integration is enabled.

   • After installation, add CUDA binaries to the system's PATH:

     1. Press Win + S, type Environment Variables, and open it.

     2. In System Properties, click Environment Variables.

     3. Under User Variables for ..., find and edit the Path variable.

     4. Click New, then add the following paths (adjust for your CUDA version): If the version installed is v12.8 (adjust according to the version installed on your machine)

        C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.8\bin
        C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.8\lib\x64
        C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.8\include

     5. Click OK to save changes.

   • Verify CUDA Installation: Windows key + R (to open the Run dialog), then type "cmd" and press Enter to open command prompt, type

     nvcc --version

     nvidia-smi

   • Restart your computer: Restart your system to apply the updates.

3. Open Anaconda Prompt
   Search for "Anaconda Prompt" in the Start menu and open it.

4. Create a New Conda Environment
   Create a new environment with Python 3.11:

   conda create -n Ambot_env python=3.11

5. Activate the Environment
   Activate the newly created environment:

   conda activate Ambot_env

6. Clone the Repository
   Clone the Ambot repository:

   git clone https://github.com/NgFEP/Ambot.git

7. Change Directory
   Navigate to the cloned repository:

   cd Ambot

8. Install Dependencies
   Install the required Python packages:

   pip install -r requirements.txt

9. Install Ollama
   install Ollama:

   conda install conda-forge::ollama

10. Pull Models
    Pull the required models using Ollama:

    ollama pull deepseek-r1:14b
    ollama pull nomic-embed-text:latest

11. Create Embeddings Folder
    Create a folder in case it's not there already to store the embeddings:

    mkdir embeddings

12. Run the Scripts
    To generate embeddings and create Ambot:

    python embedding_generation.py

    To query and test Ambot:

    python query_retrieval.py

    Ask your technical questions. Example: How do I analyze the results of an AMBER simulation using cpptraj?

Rest similar to Installation Instructions (Ubuntu)

💖 If you like this project, give it a star! ⭐

Authors

Contributors names and contact info

• Saikat Pal