NLLB-CLIP -- train performant multilingual image retrieval model on a budget
Abstract
Today, the exponential rise of large models developed by academic and industrial institutions with the help of massive computing resources raises the question of whether someone without access to such resources can make a valuable scientific contribution. To explore this, we tried to solve the challenging task of multilingual image retrieval having a limited budget of $1,000. As a result, we present NLLB-CLIP - CLIP model with a text encoder from the NLLB model. To train the model, we used an automatically created dataset of 106,246 good-quality images with captions in 201 languages derived from the LAION COCO dataset. We trained multiple models using image and text encoders of various sizes and kept different parts of the model frozen during the training. We thoroughly analyzed the trained models using existing evaluation datasets and newly created XTD200 and Flickr30k-200 datasets. We show that NLLB-CLIP is comparable in quality to state-of-the-art models and significantly outperforms them on low-resource languages.
Community
Super dangerous to Lie on Co, Co. All things Considered. Im thinking... sumbody* does both What(co?), what(co?) meaning can provide two supreme aesthetics qualities, then what? You gonna shut him in prison for doing booth good and pump numb values into what so called "scientific", without massive computing resources? I mean how stupid does it sound? And how ugly do you think you are to even consider that abstract, human!?