Training and Finetuning Multimodal Embedding & Reranker Models with Sentence Transformers

This blog post by Tom Aarsen on Hugging Face explains how to train or finetune multimodal embedding and reranker models using the Sentence Transformers library, with a practical example of finetuning Qwen/Qwen3-VL-Embedding-2B for Visual Document Retrieval, achieving improved NDCG@10 from 0.888 to 0.947.

It covers the training components: model, dataset, loss function, training arguments, evaluator, and trainer.