Scaling (Down) CLIP: A Comprehensive Analysis of Data, Architecture, and Training Strategies
It provides practical guidance for making CLIP models more accessible and affordable for various applications, though it is incremental as it builds on existing CLIP methods.
This paper analyzes how to scale down CLIP models for limited compute budgets by examining data quality, architecture choices, and training strategies, finding that high-quality data can outperform larger low-quality datasets and that CLIP+Data Augmentation achieves comparable performance with half the data.
This paper investigates the performance of the Contrastive Language-Image Pre-training (CLIP) when scaled down to limited computation budgets. We explore CLIP along three dimensions: data, architecture, and training strategies. With regards to data, we demonstrate the significance of high-quality training data and show that a smaller dataset of high-quality data can outperform a larger dataset with lower quality. We also examine how model performance varies with different dataset sizes, suggesting that smaller ViT models are better suited for smaller datasets, while larger models perform better on larger datasets with fixed compute. Additionally, we provide guidance on when to choose a CNN-based architecture or a ViT-based architecture for CLIP training. We compare four CLIP training strategies - SLIP, FLIP, CLIP, and CLIP+Data Augmentation - and show that the choice of training strategy depends on the available compute resource. Our analysis reveals that CLIP+Data Augmentation can achieve comparable performance to CLIP using only half of the training data. This work provides practical insights into how to effectively train and deploy CLIP models, making them more accessible and affordable for practical use in various applications.