Skip to main content

​
Requirements

Before installing MilesONerd AI Telegram Bot, ensure you have the following prerequisites:
  • Python 3.8 or higher - The bot requires Python 3.8+
  • pip - Python package manager (usually comes with Python)
  • Telegram Bot Token - Obtain from @BotFather on Telegram
  • Sufficient disk space - AI models require several GB of storage
  • GPU (optional) - NVIDIA GPU with CUDA support for faster inference
The bot uses Llama 3.1-Nemotron (70B) and BART models which are resource-intensive. Ensure you have adequate RAM (16GB+ recommended) and storage space.

​
Installation Steps

1

Clone the Repository

Clone the MilesONerd AI Telegram Bot repository from GitHub:
git clone https://github.com/MilesONerd/telegram-bot.git
cd telegram-bot
2

Install Python Dependencies

Install all required Python packages using pip:
pip install -r requirements.txt
The requirements.txt includes essential packages like python-telegram-bot, transformers, torch, and model dependencies.
3

Configure Environment Variables

Copy the example environment file and configure your settings:
cp .env.example .env
Edit the .env file with your credentials (see Configuration for details).
4

Verify Installation

Ensure all dependencies are correctly installed:
python -c "import telegram, transformers, torch; print('All packages installed successfully!')"

​
Key Dependencies

The bot relies on these core packages:
requirements.txt
python-telegram-bot==21.10    # Telegram Bot API wrapper
transformers==4.48.0          # Hugging Face Transformers
torch==2.5.1                  # PyTorch for model inference
python-dotenv==1.0.1          # Environment variable management
huggingface-hub==0.27.1       # Model downloading and caching

​
GPU Support

If you have an NVIDIA GPU with CUDA support, the bot will automatically detect and use it for faster inference:
The AI handler checks for CUDA availability at startup and automatically configures the models to use GPU if available. No additional configuration needed!
CUDA-related packages are included in requirements.txt: 
nvidia-cuda-runtime-cu12==12.4.127
nvidia-cudnn-cu12==9.1.0.70
triton==3.1.0

​
Troubleshooting

This means the python-telegram-bot package wasn’t installed correctly. Run:
pip install python-telegram-bot==21.10
The models are large and may exceed available GPU memory. The bot automatically falls back to CPU if GPU memory is insufficient. You can also set smaller batch sizes or use model quantization.
The first run downloads large model files from Hugging Face Hub. This is normal and only happens once. Models are cached locally for future use.
Never commit your .env file containing your bot token to version control. Keep it private and secure.

​
Next Steps

After installation, proceed to:

Build docs developers (and LLMs) love

Get started for freeTalk to us