A Practitioner's Guide to Continual Multimodal Pretraining

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Roth, Karsten, Udandarao, Vishaal, Dziadzio, Sebastian, Prabhu, Ameya, Cherti, Mehdi, Vinyals, Oriol, Hénaff, Olivier, Albanie, Samuel, Bethge, Matthias, Akata, Zeynep
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866912147014942720
author Roth, Karsten
Udandarao, Vishaal
Dziadzio, Sebastian
Prabhu, Ameya
Cherti, Mehdi
Vinyals, Oriol
Hénaff, Olivier
Albanie, Samuel
Bethge, Matthias
Akata, Zeynep
author_facet Roth, Karsten
Udandarao, Vishaal
Dziadzio, Sebastian
Prabhu, Ameya
Cherti, Mehdi
Vinyals, Oriol
Hénaff, Olivier
Albanie, Samuel
Bethge, Matthias
Akata, Zeynep
contents 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 Still, despite being pretrained on extensive data, they become outdated over time.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 often 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 as well as provide 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) A data-centric investigation of data mixtures and stream orderings that emulate real-world deployment situations, (2) a method-centric investigation 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) the influence of model and compute scaling. Together, our insights provide a practitioner's guide to continual multimodal pretraining for real-world deployment. Our benchmark and code is here: https://github.com/ExplainableML/fomo_in_flux.
format Preprint
id arxiv_https___arxiv_org_abs_2408_14471
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle A Practitioner's Guide to Continual Multimodal Pretraining
Roth, Karsten
Udandarao, Vishaal
Dziadzio, Sebastian
Prabhu, Ameya
Cherti, Mehdi
Vinyals, Oriol
Hénaff, Olivier
Albanie, Samuel
Bethge, Matthias
Akata, Zeynep
Computer Vision and Pattern Recognition
Computation and Language
Machine Learning
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 often 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 as well as provide 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) A data-centric investigation of data mixtures and stream orderings that emulate real-world deployment situations, (2) a method-centric investigation 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) the influence of model and compute scaling. Together, our insights provide a practitioner's guide to continual multimodal pretraining for real-world deployment. Our benchmark and code is here: https://github.com/ExplainableML/fomo_in_flux.
title A Practitioner's Guide to Continual Multimodal Pretraining
topic Computer Vision and Pattern Recognition
Computation and Language
Machine Learning
url https://arxiv.org/abs/2408.14471