Scaling Laws for Autoregressive Generative Modeling

by   Tom Henighan, et al.

We identify empirical scaling laws for the cross-entropy loss in four domains: generative image modeling, video modeling, multimodal image↔text models, and mathematical problem solving. In all cases autoregressive Transformers smoothly improve in performance as model size and compute budgets increase, following a power-law plus constant scaling law. The optimal model size also depends on the compute budget through a power-law, with exponents that are nearly universal across all data domains. The cross-entropy loss has an information theoretic interpretation as S(True) + D_KL(True||Model), and the empirical scaling laws suggest a prediction for both the true data distribution's entropy and the KL divergence between the true and model distributions. With this interpretation, billion-parameter Transformers are nearly perfect models of the YFCC100M image distribution downsampled to an 8× 8 resolution, and we can forecast the model size needed to achieve any given reducible loss (ie D_KL) in nats/image for other resolutions. We find a number of additional scaling laws in specific domains: (a) we identify a scaling relation for the mutual information between captions and images in multimodal models, and show how to answer the question "Is a picture worth a thousand words?"; (b) in the case of mathematical problem solving, we identify scaling laws for model performance when extrapolating beyond the training distribution; (c) we finetune generative image models for ImageNet classification and find smooth scaling of the classification loss and error rate, even as the generative loss levels off. Taken together, these results strengthen the case that scaling laws have important implications for neural network performance, including on downstream tasks.


page 11

page 24

page 25

page 29


Scaling Laws for Acoustic Models

There is a recent trend in machine learning to increase model quality by...

Learning Curve Theory

Recently a number of empirical "universal" scaling law papers have been ...

Scaling Laws for Neural Language Models

We study empirical scaling laws for language model performance on the cr...

An Optimization-Based Generative Model of Power Laws Using a New Information Theory Based Metric

In this paper, we propose an optimization-based mechanism to explain pow...

Is the Number of Trainable Parameters All That Actually Matters?

Recent work has identified simple empirical scaling laws for language mo...

Scaling Laws for Neural Machine Translation

We present an empirical study of scaling properties of encoder-decoder T...

Can machine learning identify interesting mathematics? An exploration using empirically observed laws

We explore the possibility of using machine learning to identify interes...