README.md

This repository contains experimental lora tests.

Anirand_v1_000002250

This LoRA is designed to generate anime-style images.

• Training Framework: Utilized ai-toolkit for the training process.
• Training Data: The model was trained on a small dataset of 13 images.
• Training Steps: 2250 steps were used for training.
• Learning Rate: Set to 2e-4.
• Model: Based on black-forest-labs/FLUX.1-dev
• LoRA Configuration:
  • Rank: 32
  • Batch Size: 1
  • Noise Scheduler: flowmatch
  • Resolution: Images resized to 512, 768, and 1024.
    

Anirand_v2_000002250

This LoRA is designed to generate anime-style images.

• Training Framework: Utilized ai-toolkit for the training process.
• Training Data: Same as Anirand_v1_000002250
• Training Steps: 2250 steps were used for training.
• Learning Rate: Set to 2e-4.
• Model: Based on black-forest-labs/FLUX.1-dev
• LoRA Configuration:
  • Rank: 64
  • Batch Size: 1
  • Noise Scheduler: flowmatch
  • Resolution: Images resized to 1024.