AndyLone22/MirrorMetrics
MirrorMetrics: How to evaluate Stable Diffusion LoRAs. A visual diagnostic tool to detect overfitting, check dataset quality, and fix training settings using InsightFace biometrics.
This tool helps AI artists and researchers evaluate how well their Stable Diffusion Face LoRAs maintain a consistent character identity. You input the reference images used for training and the images generated by your LoRA, and it outputs an interactive dashboard and detailed reports. This allows you to visually and numerically understand if your LoRA is overtrained, if your dataset has inconsistencies, or why generated faces might look rigid or off-model.
Use this if you train Stable Diffusion models (LoRAs) specifically for generating consistent character faces and need a scientific way to benchmark their performance, identify overfitting, or check the quality of your training data.
Not ideal if you are working with non-face image generation, general style transfer, or if you only need a quick qualitative assessment without deep biometric analysis.
Stars
48
Forks
5
Language
Python
License
MIT
Category
Last pushed
Feb 21, 2026
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/AndyLone22/MirrorMetrics"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
KohakuBlueleaf/LyCORIS
Lora beYond Conventional methods, Other Rank adaptation Implementations for Stable diffusion.
tsiendragon/qwen-image-finetune
Repo for Qwen Image Finetune
Akegarasu/lora-scripts
SD-Trainer. LoRA & Dreambooth training scripts & GUI use kohya-ss's trainer, for diffusion model.
PRITHIVSAKTHIUR/Qwen-Image-Edit-2511-LoRAs-Fast-Lazy-Load
Demonstration for the Qwen-Image-Edit-2511 model with lazy-loaded LoRA adapters for advanced...
cloneofsimo/lora
Using Low-rank adaptation to quickly fine-tune diffusion models.