Large-scale multimodal representation learning successfully optimizes for zero-shot transfer at test time. Yet the standard pretraining paradigm (contrastive learning on large amounts of image-text data) does not explicitly encourage representations to support few-shot adaptation. In this work, we propose a simple, but carefully designed extension to multimodal pretraining which enables representations to accommodate additional context. Using this objective, we show that vision-language models can be trained to exhibit significantly increased few-shot adaptation: across 21 downstream tasks, we find up to fourfold improvements in test-time sample efficiency, and average few-shot adaptation gains of over 5%, while retaining zero-shot generalization performance across model scales and training durations. In particular, equipped with simple, training-free, metric-based adaptation mechanisms, our representations easily surpass more complex and expensive optimization-based schemes, vastly simplifying generalization to new domains.

Knowledge distillation (KD) is the de facto standard for compressing large-scale multimodal models into smaller ones. Prior works have explored ever more complex KD strategies involving different objectives, teacher-ensembles, and weight inheritance. In this work, we explore an alternative, yet simple approach—active data curation as effective distillation for contrastive multimodal pretraining. Our simple online batch selection method, ACID, outperforms strong KD baselines across various model-,data-and compute-configurations. Further, we find such an active curation strategy to in fact be complementary to standard KD, and can be effectively combined to train highly performant inference-efficient models. Our simple and scalable pretraining framework, ACED, achieves state-of-the-art results across 27 zero-shot classification and image-text retrieval tasks with upto 11% less inference FLOPs. We further demonstrate that ACED yields strong vision-encoders for training generative multimodal models, outperforming larger vision encoders on image-captioning and visual question-answering tasks.

Abstract Power-law scaling indicates that large-scale training with uniform sampling is prohibitively slow. Active learning methods aim to increase data efficiency by prioritizing learning on the most relevant examples. Despite their appeal, these methods have yet to be widely adopted since no one algorithm has been shown to a) generalize across models and tasks b) scale to large datasets and c) yield overall FLOP savings when accounting for the overhead of data selection. In this work we propose a method which satisfies these three properties, leveraging small, cheap proxy models to estimate “learnability” scores for datapoints, which are used to prioritize data for training much larger models. As a result, models trained using our methods – ClassAct and ActiveCLIP – require 46% and 51% fewer training updates and up to 25% less total computation to reach the same performance as uniformly-trained visual classifiers on JFT and multimodal models on ALIGN, respectively. Finally, we find our data-prioritization scheme to be complementary with recent data-curation and learning objectives, yielding a new state-of-the-art in several multimodal transfer tasks.

Training deep networks requires various design decisions regarding for instance their architecture, data augmentation, or optimization. In this work, we find these training variations to result in networks learning unique feature sets from the data. Using public model libraries comprising thousands of models trained on canonical datasets like ImageNet, we observe that for arbitrary pairings of pretrained models, one model extracts significant data context unavailable in the other – independent of overall performance. Given any arbitrary pairing of pretrained models and no external rankings (such as separate test sets, e.g. due to data privacy), we investigate if it is possible to transfer such "complementary" knowledge from one model to another without performance degradation – a task made particularly difficult as additional knowledge can be contained in stronger, equiperformant or weaker models. Yet facilitating robust transfer in scenarios agnostic to pretrained model pairings would unlock auxiliary gains and knowledge fusion from any model repository without restrictions on model and problem specifics - including from weaker, lower-performance models. This work therefore provides an initial, in-depth exploration on the viability of such general-purpose knowledge transfer. Across large-scale experiments, we first reveal the shortcomings of standard knowledge distillation techniques, and then propose a much more general extension through data partitioning for successful transfer between nearly all pretrained models, which we show can also be done unsupervised. Finally, we assess both the scalability and impact of fundamental model properties on successful model-agnostic knowledge transfer.

Most transformer-based video encoders are limited to short temporal contexts due to their quadratic complexity. While various attempts have been made to extend this context, this has often come at the cost of both conceptual and computational complexity. We propose to instead re-purpose existing pre-trained video transformers by simply fine-tuning them to attend to memories derived non-parametrically from past activations. By leveraging redundancy reduction, our memory-consolidated vision transformer (MC-ViT) effortlessly extends its context far into the past and exhibits excellent scaling behavior when learning from longer videos. In doing so, MC-ViT sets a new state-of-the-art in long-context video understanding on EgoSchema, Perception Test, and Diving48, outperforming methods that benefit from orders of magnitude more parameters.

Data curation is an essential component of large-scale pretraining. In this work, we demonstrate that jointly prioritizing batches of data is more effective for learning than selecting examples independently. Multimodal contrastive objectives expose the dependencies between data and thus naturally yield criteria for measuring the joint learnability of a batch. We derive a simple and tractable algorithm for selecting such batches, which significantly accelerate training beyond individually-prioritized data points. As performance improves by selecting from large super-batches, we also leverage recent advances in model approximation to reduce the computational overhead of scoring. As a result, our approach—multimodal contrastive learning with joint example selection (JEST)—surpasses state-of-the-art pretraining methods with up to 13× fewer iterations and 10× less computation. Essential to the performance of JEST is the ability to steer the data selection process towards the distribution of smaller, well-curated datasets via pretrained reference models, exposing data curation as a new dimension for neural scaling laws.

Multimodal foundation models serve numerous applications at the intersection of vision and language. Still, despite being pretrained on extensive data, they become outdated over time.To keep models updated, research into continual pretraining mainly explores scenarios with either (1) infrequent, indiscriminate updates on large-scale new data, or (2) frequent, sample-level updates.However, practical model deployment often operates in the gap between these two limit cases, as real-world applications demand adaptation to specific subdomains, tasks or concepts --- spread over the entire, varying life cycle of a model. In this work, we complement current perspectives on continual pretraining through a research test bed and offer comprehensive guidance for effective continual model updates in such scenarios.We first introduce FoMo-in-Flux, a continual multimodal pretraining benchmark with realistic compute constraints and practical deployment requirements, constructed over 63 datasets with diverse visual and semantic coverage.Using FoMo-in-Flux, we explore the complex landscape of practical continual pretraining through multiple perspectives: (1) data mixtures and stream orderings that emulate real-world deployment settings, (2) methods ranging from simple fine-tuning and traditional continual learning strategies to parameter-efficient updates and model merging, (3) meta-learning-rate schedules and mechanistic design choices, and (4) model and compute scaling. Together, our insights provide a practitioner's guide to continual multimodal pretraining for real-world deployment. Benchmark and code is provided here: https://github.com/ExplainableML/fomoinflux.

In-context learning––the ability to configure a model's behavior with different prompts––has revolutionized the field of natural language processing, alleviating the need for task-specific models and paving the way for generalist models capable of assisting with any query. Computer vision, in contrast, has largely stayed in the former regime: specialized decoders and finetuning protocols are generally required to perform dense tasks such as semantic segmentation and depth estimation. In this work we explore a simple mechanism for in-context learning of such scene understanding tasks: nearest neighbor retrieval from a prompt of annotated features. We propose a new pretraining protocol––leveraging attention within and across images––which yields representations particularly useful in this regime. The resulting Hummingbird model, suitably prompted, performs various scene understanding tasks without modification while approaching the performance of specialists that have been finetuned for each task. Moreover, Hummingbird can be configured to perform new tasks much more efficiently than finetuned models, raising the possibility of scene understanding in the interactive assistant regime.

Humans learn powerful representations of objects and scenes by observing how they evolve over time. Yet, outside of specific tasks that require explicit temporal understanding, static image pretraining remains the dominant paradigm for learning visual foundation models. We question this mismatch, and ask whether video pretraining can yield visual representations that bear the hallmarks of human perception: generalisation across tasks, robustness to perturbations, and consistency with human judgements. To that end we propose a novel procedure for curating videos, and develop a contrastive framework which learns from the complex transformations therein. This simple paradigm for distilling knowledge from videos, called VITO, yields general representations that far outperform prior video pretraining methods on image understanding tasks, and image pretraining methods on video understanding tasks. Moreover, VITO representations are significantly more robust to natural and synthetic deformations than image-, video-, and adversarially-trained ones. Finally, VITO’s predictions are strongly aligned with human judgements, surpassing models that were specifically trained for that purpose. Together, these results suggest that video pretraining could be a simple way of learning unified, robust, and human-aligned representations of the visual world.

Self-supervised pretraining has been shown to yield powerful representations for transfer learning. These performance gains come at a large computational cost however, with state-of-the-art methods requiring an order of magnitude more computation than supervised pretraining. We tackle this computational bottleneck by introducing a new self-supervised objective, contrastive detection, which tasks representations with identifying object-level features across augmentations. This objective extracts a rich learning signal per image, leading to state-of-the-art transfer accuracy on a variety of downstream tasks, while requiring up to 10× less pretraining. In particular, our strongest ImageNet-pretrained model performs on par with SEER, one of the largest self-supervised systems to date, which uses 1000× more pretraining data. Finally, our objective seamlessly handles pretraining on more complex images such as those in COCO, closing the gap with supervised transfer learning from COCO to PASCAL.

Self-supervised learning holds promise in leveraging large amounts of unlabeled data, however much of its progress has thus far been limited to highly curated pre-training data such as ImageNet. We explore the effects of contrastive learning from larger, less-curated image datasets such as YFCC, and find there is indeed a large difference in the resulting representation quality. We hypothesize that this curation gap is due to a shift in the distribution of image classes—which is more diverse and heavy-tailed—resulting in less relevant negative samples to learn from. We test this hypothesis with a new approach, Divide and Contrast (DnC), which alternates between contrastive learning and clustering-based hard negative mining. When pretrained on less curated datasets, DnC greatly improves the performance of self-supervised learning on downstream tasks, while remaining competitive with the current state-of-the-art on curated datasets.

Human observers can learn to recognize new categories of images from a handful of examples, yet doing so with artificial ones remains an open challenge. We hypothesize that data-efficient recognition is enabled by representations which make the variability in natural signals more predictable. We therefore revisit and improve Contrastive Predictive Coding, an unsupervised objective for learning such representations. This new implementation produces features which support state-of-the-art linear classification accuracy on the ImageNet dataset. When used as input for non-linear classification with deep neural networks, this representation allows us to use 2-5x less labels than classifiers trained directly on image pixels. Finally, this unsupervised representation substantially improves transfer learning to object detection on the PASCAL VOC dataset, surpassing fully supervised pre-trained ImageNet classifiers.