This is a one click installer to set up a local Telegram bot powered by a language model. It will:
- Clone the llama-telegram-bot repository
- Install the requirements
- Download the WizardLM model
- Set the required environment variables
- Run the Telegram bot
- Windows system
- Git installed
- Python 3.6+
You can obtain a Telegram bot token in less than 30 seconds easily by messaging @BotFather on Telegram and following the steps to create a new bot.
- Double click the
install.bat
file. - The bot will start installing in your terminal.
- Double click the
Start.bat
file. - The bot will start running in your terminal.
- Message the bot you created with BotFather
@YourBotName
on Telegram to start chatting!
You can customize the following:
In install.bat
:
BOT_TOKEN
- The token for your Telegram botMODEL_URL
- The URL to download a different language modelALLOWED_USERS
- A space separated list of Telegram usernames allowed to access the bot
In start.bat
:
BOT_TOKEN
- The token for your Telegram botMODEL_PATH
- The path to your downloaded modelALLOWED_USERS
- A space separated list of Telegram usernames allowed to access the bot
This project uses:
- llama-telegram-bot for the Telegram bot
- Hugging Face to host the language model
- python-telegram-bot library to interface with the Telegram API
- llama.cpp for the language model
- llama-cpp-python for the language model
- The default model used is a Wizard101-themed model hosted on Hugging Face 🧙
- This bot is for entertainment purposes only.
- The owner of this bot assumes no responsibility for inappropriate or offensive responses from the AI.
Let me know if you have any other questions!