🦙 Ollama Telegram Bot

Chat with your LLM, using Telegram bot!
Feel free to contribute!

Features

Here's features that you get out of the box:

• Fully dockerized bot
• Response streaming without ratelimit with SentenceBySentence method
• Mention [@] bot in group to receive answer

Roadmap

• Proper Docker Config
• Add more API-related functions [System Prompt Editor, Ollama Version fetcher, etc.]
• Redis DB integration
• Implement history [Bot can't remember more that 1 prompt]

Prerequisites

• Telegram-Bot Token

Installation (Non-Docker)

• Install latest Python
• Clone Repository

git clone https://github.com/ruecat/ollama-telegram

• Install requirements from requirements.txt

pip install -r requirements.txt

• Enter all values in .env.example

• Rename .env.example -> .env

• Launch bot

python3 run.py

Installation (Docker-Compose)

• Clone Repository

git clone https://github.com/ruecat/ollama-telegram

• Enter all values in .env.example

• Rename .env.example -> .env

• Run ONE of the following docker compose commands to start:

  1. To run ollama in docker container (optionally: uncomment GPU part of docker-compose.yml file to enable Nvidia GPU)

  docker compose up --build -d
  

  2. To run ollama from locally installed instance (mainly for MacOS, since docker image doesn't support Apple GPU acceleration yet):

  docker compose up --build -d ollama-telegram
  

Environment Configuration

Parameter	Description	Required?	Default Value	Example
TOKEN	Your Telegram bot token. [How to get token?]	Yes	yourtoken	MTA0M****.GY5L5F.****g*****5k
ADMIN_IDS	Telegram user IDs of admins. These can change model and control the bot.	Yes		1234567890 OR 1234567890,0987654321, etc.
USER_IDS	Telegram user IDs of regular users. These only can chat with the bot.	Yes		1234567890 OR 1234567890,0987654321, etc.
INITMODEL	Default LLM	No	llama2	mistral:latest mistral:7b-instruct

Credits

• Ollama

Libraries used

• Aiogram 3.x