State-of-the-art CLIP/SigLIP embedding models finetuned for the fashion domain. +57% increase in evaluation metrics vs FashionCLIP 2.0.

Fine-Tuning SigLIP 2 for Single/Multi-Label Image Classification. Image classification vision-language encoder model fine-tuned for Image Classification Tasks

A comprehensive fine-tuning framework for image classification using Google's Siglip2 vision model with advanced training strategies including progressive unfreezing, weighted sampling for imbalanced datasets, and extensive configuration options.

No description available