Evil Moe Discord Bot (Docker Compose Setup)

This guide explains how to set up and run the emoe aka Evil Moe Discord Bot using Docker and Docker Compose. This method bundles the bot and its Ollama dependency into manageable containers.If you cant host the bot on your own and had errors let me know in discord my id is idc.btw also if you want the bot and doesnt have vps , dm i maybe able to help with my vps

here is some examples of chat

Prerequisites

• VPS or A local machine : it can be linux / windows / mac : takes around 4gb of ram and 6gb of storage . No gpu is needed a cpu with 2 core around 2.5 ghz will be enough and fast.
• Git: To clone the repository.
• Docker & Docker Compose: Required to build and run the containerized application.
• Discord Bot Token: A secret token needed for the bot to connect to Discord.

Getting a Discord Bot Token

1. Go to the Discord Developer Portal.
2. Click "New Application" and give your bot a name.
3. Navigate to the "Bot" tab on the left sidebar.
4. Click "Add Bot" and confirm.
5. Under the bot's username, click "Reset Token" (or "View Token" if visible) and copy the token. Treat this token like a password - do not share it!
6. While you're here, ensure the following Privileged Gateway Intents are enabled (required for the bot to read messages):
   • PRESENCE INTENT (May not be strictly needed for this bot, but often useful)
   • SERVER MEMBERS INTENT (May not be strictly needed for this bot, but often useful)
   • MESSAGE CONTENT INTENT (Required)
7. You also need to invite the bot to your server. Go to the "OAuth2" -> "URL Generator" tab. Select the bot and application.commands scopes. Then, select the necessary Bot Permissions (e.g., Read Messages/View Channels, Send Messages, Embed Links, Attach Files). Copy the generated URL and paste it into your browser to invite the bot to your desired server.

Installation Steps

1. Install Docker and Docker Compose

You need both Docker Engine and Docker Compose. Installation varies by operating system:

• Windows: Install Docker Desktop for Windows. It includes both Docker Engine and Docker Compose. Follow the official instructions: https://docs.docker.com/desktop/install/windows-install/
• macOS: Install Docker Desktop for Mac. It includes both Docker Engine and Docker Compose. Follow the official instructions: https://docs.docker.com/desktop/install/mac-install/
• Linux:
  • Install Docker Engine for your specific distribution. Follow the official instructions: https://docs.docker.com/engine/install/#server
  • Install the Docker Compose Plugin (recommended) or standalone docker-compose. Follow the official instructions: https://docs.docker.com/compose/install/linux/
  • Ensure you complete the post-installation steps for Linux to run Docker as a non-root user: https://docs.docker.com/engine/install/linux-postinstall/

Verify your installation by running docker --version and docker compose version (or docker-compose --version for older standalone installs) in your terminal.

2. Clone the Repository

Open your terminal or command prompt and clone the repository:

git clone https://github.com/aptdnfapt/emoe.git

3. Navigate to the Directory

Change into the cloned repository's directory:

cd emoe

4. Configure .env File (Manual Token Input)

The installation script will prompt you for your Discord Bot Token. Make sure you have it ready from the Prerequisites step. The script will automatically update the .env file.

(See the "Configuration Details" section below for explanations of other .env variables like replacements).

5. Run the Installation Script

Execute the installation script. This will guide you through the setup process, including entering your Discord Bot Token and downloading the necessary Ollama model.

./install.sh

(If you get a permission error, run chmod +x install.sh first)

The script will:

• Check for Docker and Docker Compose.
• Prompt you for your Discord Bot Token and update the .env file.
• Create necessary placeholder files (bot_config.json, chat_log.jsonl).
• Start the Ollama service via Docker Compose.
• Download the required Ollama model (hf.co/mradermacher/DialoGPT-large-gavin-GGUF:F16) inside the Ollama container (this may take time).
• Build the bot's Docker image (if needed).
• Start both the bot and Ollama containers.

INSTALL emoe Natively on Ubuntu 22.04 (Without Docker) (advance users)

This section describes how to run the bot directly on an Ubuntu 22.04 system without using Docker.

Prerequisites (Native Install)

• Ubuntu 22.04 System: With user access (sudo privileges may be needed for installations) but you can go with any other linux distro the pkg manager will be different .
• Git: To clone the repository (sudo apt update && sudo apt install git -y).
• Python 3.10 or later: Ubuntu 22.04 usually comes with a compatible version. Check with python3 --version. You'll also need pip and venv. Install if missing: sudo apt update && sudo apt install python3 python3-pip python3-venv -y.
• Ollama: Installed locally on the system.
• tmux or screen (Recommended): To keep the bot running after you disconnect (sudo apt update && sudo apt install tmux -y).
• Discord Bot Token: See "Getting a Discord Bot Token" section above.

1. Install Ollama Locally

Follow the official Ollama instructions to install it on your Linux system. Usually, this involves a command like:

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

Verify the installation by running ollama --version.

2. Pull the Required Model

Once Ollama is installed, pull the specific model used by the bot:

ollama pull hf.co/mradermacher/DialoGPT-large-gavin-GGUF:F16

Wait for the download to complete. You can check available models with ollama list. Ensure Ollama is running (it usually starts automatically as a service).

3. Clone the Repository and Navigate

git clone https://github.com/aptdnfapt/emoe.git
cd emoe

4. Set Up Python Virtual Environment(optional)

It's recommended to use a virtual environment to manage dependencies and quickly remove the bot if you dont like it . but the pkgs are not soo huge so u can skip it too :

python3 -m venv venv
source venv/bin/activate

Your terminal prompt should now show (venv).

5. Install Python Dependencies

pip install -r requirements.txt

6. Configure .env File

Copy the example environment file and edit it:

nano .env # Or use your preferred text editor

• Set DISCORD_BOT_TOKEN to your actual bot token.
• Crucially, change OLLAMA_API_URL to point to your local Ollama instance. The default is usually http://localhost:11434. So, ensure the line looks like:

  OLLAMA_API_URL=http://127.0.0.1:11434
  

7. Run the Bot

Make sure your virtual environment is active (source venv/bin/activate if needed). Then, run the bot script:

python moe_bot.py

The bot should start, connect to Discord, and print "Emoe Bot is ready."

8. Keep the Bot Running (Recommended: tmux)

To keep the bot running after you close your terminal:

1. Start a new tmux session: tmux new -s moe
2. Activate the virtual environment: source venv/bin/activate
3. Run the bot: python moe_bot.py
4. Detach from the tmux session: Press Ctrl+b then d.

The bot will continue running in the background. You can reattach to the session later using tmux attach -t moe.

How it Works -- {DOCKER}

• Docker Compose (docker-compose.yml): Defines and manages two services: ollama (running the official Ollama image) and bot (built using the Dockerfile). It sets up networking so they can communicate and configures automatic restarts (restart: unless-stopped).
• Dockerfile: Specifies how to build the Docker image for the Python bot, including installing dependencies from requirements.txt.
• .env File: Stores configuration variables like your DISCORD_BOT_TOKEN and the OLLAMA_API_URL (which points to the Ollama service within the Docker network).
• Volumes: Docker Compose uses volumes to persist data:
  • ollama_data: Stores the downloaded Ollama models, so they aren't lost when the container stops.
  • ./bot_config.json & ./chat_log.jsonl: Links files inside the bot container to files in your emoe directory on your host machine, making the bot's configuration and logs persistent.
• install.sh: Automates the setup steps: prerequisite checks, .env configuration, placeholder file creation, Ollama model download, and starting the services via docker-compose up -d.
• uninstall.sh: Automates cleanup: stops and removes containers, removes the custom bot image, removes the Ollama data volume, removes the base Ollama image, and removes the host config/log files. /

Configuration Details (.env File)

The .env file controls the bot's core settings and customization.

• DISCORD_BOT_TOKEN: (Required) Your unique bot token obtained from the Discord Developer Portal. The install.sh script will prompt you for this.
• OLLAMA_API_URL: (Required for Docker Compose) Should be set to http://ollama:11434 to allow the bot container to communicate with the Ollama container service. Do not change this unless you have a custom Docker network setup.
• WORD_R/EPLACEMENTS: (Optional) Define standard, case-insensitive word or phrase replacements that always happen.
  • Format: Comma-separated pairs of word_to_find:replacement_word. No extra spaces around : or ,.
  • Example: WORD_REPLACEMENTS=gaven:moe,hello there:greetings
• PROBABILISTIC_REPLACEMENTS: (Optional) Define case-insensitive replacements that happen based on chance.
  • Format: Semicolon-separated rules. Each rule is target|probability|option1,option2,....
    • target: The word/phrase/emoji to find.
    • probability: A number from 0.0 (never) to 1.0 (always) representing the chance of replacement for each occurrence found.
    • option1,option2,...: Comma-separated list of possible replacements. One is chosen randomly if the probability check succeeds.
  • Example: PROBABILISTIC_REPLACEMENTS=💀|0.5|🍆,😘,🦅;👍|0.7|💀 (Replaces 💀 with 🍆, 😘, or 🦅 50% of the time; replaces 👍 with 💀 70% of the time).

Managing the Bot

• View Logs: docker-compose logs -f bot or docker-compose logs -f ollama

• Stop Services: docker-compose down (run in the emoe directory)

• Start Services: docker-compose up -d (run in the emoe directory)

• Update Configuration (.env):

  1. docker-compose down
  2. Edit the .env file.
  3. docker-compose up -d

• Uninstall: Run ./uninstall.sh from the emoe directory.

• Anti-Spam Cooldown: The bot has a built-in per-user cooldown of 5 seconds to prevent it from responding too rapidly to messages from the same user. Messages sent by a user within this cooldown period will be ignored.

Managing Dedicated Channels (Per Server)

The bot can be configured to respond in specific dedicated channels within each server. These commands require administrator permissions in the Discord server and must be run within that server.

• emoe add #channel-name: Adds the mentioned channel to the list of dedicated channels for this server. The bot will respond to messages in this channel without requiring trigger words.
• emoe remove #channel-name: Removes the mentioned channel from the list of dedicated channels for this server. The bot will no longer automatically respond in this channel unless messages contain trigger words.
• emoe list: Lists all channels currently configured as dedicated channels for this server.

hf repo

https://huggingface.co/mradermacher/DialoGPT-large-gavin-GGUF

Special Thanks:

Emoe was inspired by https://github.com/monk-afk/moe based and made using same model but fine-tuned dialogpt by mradermacher . I call it evil moe because the main model Dialogpt was trained by microsoft on reddit data . But still it doesnt have the reddit bs and caos. Emoe's base model aka DialoGPT-large-gavin was finned tuned to be more like that fashion . And I can promise you that it nailed it . Enjoy .

timewaste

Check out more proj? like https://github.com/aptdnfapt/OliveOwl.git