Timezone: »

Data Determines Distributional Robustness in Contrastive Language Image Pre-training (CLIP)
Alex Fang · Gabriel Ilharco · Mitchell Wortsman · Yuhao Wan · Vaishaal Shankar · Achal Dave · Ludwig Schmidt

Wed Jul 20 11:10 AM -- 11:15 AM (PDT) @ Room 318 - 320

Contrastively trained language-image models such as CLIP, ALIGN, and BASIC have demonstrated unprecedented robustness to multiple challenging natural distribution shifts. Since these language-image models differ from previous training approaches in several ways, an important question is what causes the large robustness gains. We answer this question via a systematic experimental investigation. Concretely, we study five different possible causes for the robustness gains: (i) the training set size, (ii) the training distribution, (iii) language supervision at training time, (iv) language supervision at test time, and (v) the contrastive loss function. Our experiments show that the more diverse training distribution is the main cause for the robustness gains, with the other factors contributing little to no robustness. Beyond our experimental results, we also introduce ImageNet-Captions, a version of ImageNet with original text annotations from Flickr, to enable further controlled experiments of language-image training.

Author Information

Alex Fang (University of Washington)
Gabriel Ilharco (University of Washington)
Mitchell Wortsman (University of Washington)
Yuhao Wan (University of Washington, Seattle)
Vaishaal Shankar (Amazon)
Achal Dave (Carnegie Mellon University)
Ludwig Schmidt (University of Washington)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors