*Equal Contributors

Contrastive pretraining of image-text foundation models, such as CLIP, demonstrated excellent zero-shot performance and improved robustness on a wide range of downstream tasks. However, these models utilize large transformer-based encoders with significant memory and latency overhead which pose challenges for deployment on mobile devices. In this work, we introduce MobileCLIP -- a new family of efficient image-text models optimized for runtime performance along with a novel and efficient training approach, namely multi-modal reinforced training. The proposed training approach leverages knowledge transfer from an image captioning model and an ensemble of strong CLIP encoders to improve the accuracy of efficient models. Our approach avoids train-time compute overhead by storing the additional knowledge in a reinforced dataset. MobileCLIP sets a new state-of-the-art latency-accuracy tradeoff for zero-shot classification and retrieval tasks on several datasets. Our MobileCLIP-S2 variant is 2.3 faster while more accurate compared to previous best CLIP model based on ViT-B/16. We further demonstrate the effectiveness of our multi-modal reinforced training by training a CLIP model based on ViT-B/16 image backbone and achieving +2.9% average performance improvement on 38 evaluation benchmarks compared to the previous best. Moreover, we show that the proposed approach achieves 10-1000 improved learning efficiency when compared with non-reinforced CLIP training.

Related readings and updates.

MOFI: Learning Image Representation from Noisy Entity Annotated Images

In this paper, we introduce a novel approach to automatically assign entity labels to images from existing noisy image-text pairs. The approach employees a named entity recognition model to extract entities from text, and uses a CLIP model to select the right entities as the labels of the paired image. The approach is simple, and can be readily scaled up to billions of image-text pairs mined from the web, through which we have successfully…
See paper details

Self Supervision Does Not Help Natural Language Supervision at Scale

Self supervision and natural language supervision have emerged as two exciting ways to train general purpose image encoders which excel at a variety of downstream tasks. Recent works such as M3AE [31] and SLIP [64] have suggested that these approaches can be effectively combined, but most notably their results use small (<20M examples) pre-training datasets and don’t effectively reflect the large-scale regime (>100M samples) that is commonly…
See paper details