Multimodal contrastive pretraining has been used to train multimodal representation models,
such as CLIP, on large amounts of paired image-text data. However, previous studies have …

Machine learning systems are known to be sensitive to spurious correlations between non-
essential features of the inputs (eg, background, texture, and secondary objects) and the …

While deep learning models have shown remarkable performance in various tasks, they are
susceptible to learning non-generalizable _spurious features_ rather than the core features …

The recent advance in vision-language models is largely attributed to the abundance of
image-text data. We aim to replicate this success for video-language models but there …

Abstract Vision-Language Models (VLMs) are pretrained on large diverse and noisy web-
crawled datasets. This underscores the critical need for dataset pruning as the quality of …

Fine-tuning pre-trained vision-language models, like CLIP, has yielded success on diverse
downstream tasks. However, several pain points persist for this paradigm:(i) directly tuning …

Large vision language models, such as CLIP, demonstrate impressive robustness to
spurious features than single-modal models trained on ImageNet. However, existing test …

Due to the limited availability of data, existing few-shot learning methods trained from
scratch fail to achieve satisfactory performance. In contrast, large-scale pre-trained models …

Machine learning models fail catastrophically under distribution shift, but a surprisingly
effective way to empirically improve robustness to some types of shift (* eg*, Imagenet-A/C) …

Zero-shot inference is a powerful paradigm that enables the use of large pretrained models
for downstream classification tasks without further training. However, these models are …