FastCLIP: A Suite of Optimization Techniques to Accelerate CLIP Training with Limited Resources

TL;DR

FastCLIP introduces a suite of optimization techniques tailored for resource-limited CLIP training, significantly reducing computational requirements while maintaining performance, through advanced compositional optimization, efficient communication strategies, and optimized training schedules.

Contribution

The paper presents FastCLIP, a novel training framework that enhances CLIP training efficiency on limited resources by integrating advanced optimization techniques and communication strategies.

Findings

FastCLIP achieves comparable or better performance than state-of-the-art methods on limited GPU setups.

The optimized training schedule and parameter update rules significantly improve training speed.

FastCLIP demonstrates scalability across different data sizes and compute resources.

Abstract

Existing studies of training state-of-the-art Contrastive Language-Image Pretraining (CLIP) models on large-scale data involve hundreds of or even thousands of GPUs due to the requirement of a large batch size. However, such a large amount of resources is not accessible to most people. While advanced compositional optimization techniques for optimizing global contrastive losses have been demonstrated effective for removing the requirement of large batch size, their performance on large-scale data remains underexplored and not optimized. To bridge the gap, this paper explores several aspects of CLIP training with limited resources (e.g., up to tens of GPUs). First, we introduce FastCLIP, a general CLIP training framework built on advanced compositional optimization techniques while designed and optimized for the distributed setting. Our framework is equipped with an efficient gradient reduction strategy to reduce communication overhead. Second, to further boost training efficiency, we investigate three components of the framework from an optimization perspective: the schedule of the inner learning rate, the update rules of the temperature parameter and the model parameters, respectively. Experiments on different strategies for each component shed light on how to conduct CLIP training more efficiently. Finally, we benchmark the performance of FastCLIP and the state-of-the-art training baseline (OpenCLIP) on different compute scales up to 32 GPUs on 8 nodes, and three data scales ranging from 2.7 million, 9.1 million to 315 million image-text pairs to demonstrate the significant improvement of FastCLIP in the resource-limited setting. We release the code of FastCLIP at https://github.com/Optimization-AI/fast_clip .