Are we pretraining it right? Digging deeper into visio-linguistic pretraining
This work addresses dataset selection for vision-language pretraining, revealing critical gaps in current practices for researchers in multimodal AI.
The study investigates the impact of pretraining dataset domain similarity on downstream vision-language tasks, finding that automatically generated data closer to the target domain outperforms natural data from slightly different domains, and achieves near state-of-the-art results with simple design choices.
Numerous recent works have proposed pretraining generic visio-linguistic representations and then finetuning them for downstream vision and language tasks. While architecture and objective function design choices have received attention, the choice of pretraining datasets has received little attention. In this work, we question some of the default choices made in literature. For instance, we systematically study how varying similarity between the pretraining dataset domain (textual and visual) and the downstream domain affects performance. Surprisingly, we show that automatically generated data in a domain closer to the downstream task (e.g., VQA v2) is a better choice for pretraining than "natural" data but of a slightly different domain (e.g., Conceptual Captions). On the other hand, some seemingly reasonable choices of pretraining datasets were found to be entirely ineffective for some downstream tasks. This suggests that despite the numerous recent efforts, vision & language pretraining does not quite work "out of the box" yet. Overall, as a by-product of our study, we find that simple design choices in pretraining can help us achieve close to state-of-art results on downstream tasks without any architectural changes.