HearmemanAI LoRA Trainer - Quick Start Guide

Deploy here: https://get.runpod.io/diffusion-pipe-template

Getting Started

Step 1: Open Terminal

Click the Terminal button to open a command prompt.

Step 2: Start the Interactive Training Script

Type the following command and press Enter:

bash interactive_start_training.sh

Step 3: Follow the Interactive Setup

The script will guide you through:

1. Model Selection - Choose from Flux, SDXL, or Wan models
2. API Keys - Enter required tokens (Hugging Face for Flux, Gemini for video captioning)
3. Dataset Options - Select image captioning, video captioning, or both
4. Configuration Review - Review training parameters before starting

Step 4: Wait for Training to Complete

The script will automatically:

• Download the selected model
• Generate captions for your media (if selected)
• Start LoRA training with optimized settings

Training Results

Once training is complete, your trained LoRA files will be saved in:

training_outputs

Dataset Preparation

Before running the script, place your training data in:

• Images: image_dataset_here/ folder
• Videos: video_dataset_here/ folder

Tips

• First Run: Allow extra time for model downloads (can be several GB)
• API Keys: Have your Hugging Face token ready for Flux, Gemini API key for video captioning
• Monitor Progress: The script shows progress indicators for downloads and captioning
• Review Captions: You'll be prompted to manually review generated captions before training starts

Need Help?

The interactive script provides clear instructions and error messages to guide you through each step. Simply follow the on-screen prompts!