kink-visuals

Kink Description Generator

This project helps you generate detailed descriptions for kinks/activities using local LLMs through Ollama. It processes a CSV file containing kink information and generates structured responses with consistent formatting.

Features

• Fully automated processing using local LLMs via Ollama
• Progress tracking to resume where you left off
• Configurable batch processing and delays
• Markdown output for easy reading and sharing

Setup

1. Clone the repository:

   git clone https://github.com/drourke/kink-visuals.git
   cd kink-visuals
   

2. Install dependencies:

   npm install
   

3. Make sure you have Ollama installed and running:

   # Install Ollama (if not already installed)
   # See https://github.com/ollama/ollama for installation instructions
   
   # Start Ollama server
   ollama serve
   
   # Pull the deepseek-r1:70b model (or any other model you want to use)
   ollama pull deepseek-r1:70b
   

Using Ollama

Installing Ollama

1. Visit the Ollama GitHub repository for installation instructions.
2. For macOS, you can download the app from the Ollama website.
3. For Linux, you can use the install script:

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

4. For Windows, you can use WSL or Docker.

Starting Ollama

1. If you installed the macOS app, simply open the Ollama application.
2. If you installed via command line, start the Ollama server:

   ollama serve
   

Checking Ollama Status

You can check if Ollama is running and if the required model is available:

npm run check

If the model is not available, you’ll need to pull it:

ollama pull deepseek-r1:70b

Using Different Models

The script supports using different models with Ollama. By default, it uses deepseek-r1:70b, but you can use any model available in Ollama.

Check available models:

npm run check

Use a specific model:

npm run start -- --model llama3

We’ve included shortcuts for some popular models:

• Llama 3:

  npm run llama3
  npm run llama3:one  # Process just one kink
  npm run check:llama3  # Check if the model is available
  

• Mistral:

  npm run mistral
  npm run mistral:one  # Process just one kink
  npm run check:mistral  # Check if the model is available
  

Usage

Basic Usage

Run the script to process all kinks:

npm start

Process a Single Kink

To process just one kink:

npm run one

Process a Batch of Kinks

To process a batch of kinks (default: 5) with a delay between each:

npm run batch

Fully Automated Processing

To process all kinks without prompting after each one:

npm run auto

To process a batch of kinks automatically:

npm run auto-batch

Force Start from Beginning

To force the script to start from the beginning, ignoring progress:

npm run force

To process just one kink, starting from the beginning:

npm run force-one

Custom Processing

You can specify the number of kinks to process, the delay between them, and whether to skip prompts:

node index.js --count 10 --delay 15 --no-prompt --force-start --model llama3

Command Line Options

• --count <number>: Number of kinks to process before stopping
• --delay <seconds>: Delay in seconds between processing each kink
• --no-prompt: Skip prompting after each kink (fully automated mode)
• --force-start: Force start from the beginning, ignoring progress
• --debug: Enable debug mode with verbose logging
• --model <name>: Specify which model to use

Output

All generated prompts and responses are saved in the kink_descriptions directory:

• Prompts are saved as [kink_id]_prompt.txt
• Responses are saved as [kink_id].md

Progress Tracking

Your progress is saved in progress.json. If you stop the script and run it again later, it will continue from where you left off.

How It Works

1. The script reads the CSV file containing kink information
2. For each kink, it generates a prompt based on the kink’s name, description, and suggested categories
3. It calls the Ollama API with the prompt
4. The API response is formatted as Markdown and saved to a file
5. Progress is tracked so you can stop and resume later

Requirements

• Node.js 16.x or higher
• Ollama installed and running with the deepseek-r1:70b model (or another model of your choice)

Troubleshooting

Ollama Issues

• Ollama not running: Make sure Ollama is running by starting it with ollama serve or opening the Ollama app.
• Model not found: Pull the model with ollama pull deepseek-r1:70b.
• Connection refused: Check if Ollama is running on the default port (11434).
• Slow responses: Large models like deepseek-r1:70b require significant resources. Consider using a smaller model if performance is an issue.