Scaling Laws for Sparsely-Connected Foundation Models

• URL: http://arxiv.org/abs/2309.08520v1
• Date: Fri, 15 Sep 2023 16:29:27 GMT
• Title: Scaling Laws for Sparsely-Connected Foundation Models
• Authors: Elias Frantar, Carlos Riquelme, Neil Houlsby, Dan Alistarh, Utku Evci
• Abstract summary: We explore the impact of parameter sparsity on the scaling behavior of Transformers trained on massive datasets. We identify the first scaling law describing the relationship between weight sparsity, number of non-zero parameters, and amount of training data.
• Score: 70.41266138010657
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We explore the impact of parameter sparsity on the scaling behavior of Transformers trained on massive datasets (i.e., "foundation models"), in both vision and language domains. In this setting, we identify the first scaling law describing the relationship between weight sparsity, number of non-zero parameters, and amount of training data, which we validate empirically across model and data scales; on ViT/JFT-4B and T5/C4. These results allow us to characterize the "optimal sparsity", the sparsity level which yields the best performance for a given effective model size and training budget. For a fixed number of non-zero parameters, we identify that the optimal sparsity increases with the amount of data used for training. We also extend our study to different sparsity structures (such as the hardware-friendly n:m pattern) and strategies (such as starting from a pretrained dense model). Our findings shed light on the power and limitations of weight sparsity across various parameter and computational settings, offering both theoretical understanding and practical implications for leveraging sparsity towards computational efficiency improvements.

Related papers

• How Does Data Diversity Shape the Weight Landscape of Neural Networks? [2.89287673224661]
  We investigate the impact of dropout, weight decay, and noise augmentation on the parameter space of neural networks. We observe that diverse data influences the weight landscape in a similar fashion as dropout. We conclude that synthetic data can bring more diversity into real input data, resulting in a better performance on out-of-distribution test instances.
  arXiv  Detail & Related papers  (2024-10-18T16:57:05Z)
• AutoScale: Automatic Prediction of Compute-optimal Data Composition for Training LLMs [61.13296177652599]
  This paper demonstrates that the optimal composition of training data from different domains is scale-dependent. We introduce *AutoScale*, a novel, practical approach for optimizing data compositions at potentially large training data scales. Our evaluation on GPT-2 Large and BERT pre-training demonstrates *AutoScale*'s effectiveness in improving training convergence and downstream performance.
  arXiv  Detail & Related papers  (2024-07-29T17:06:30Z)
• A Solvable Model of Neural Scaling Laws [72.8349503901712]
  Large language models with a huge number of parameters, when trained on near internet-sized number of tokens, have been empirically shown to obey neural scaling laws. We propose a statistical model -- a joint generative data model and random feature model -- that captures this neural scaling phenomenology. Key findings are the manner in which the power laws that occur in the statistics of natural datasets are extended by nonlinear random feature maps.
  arXiv  Detail & Related papers  (2022-10-30T15:13:18Z)
• Parameter-Efficient Sparsity for Large Language Models Fine-Tuning [63.321205487234074]
  We propose a. sparse-efficient Sparse Training (PST) method to reduce the number of trainable parameters during sparse-aware training. Experiments with diverse networks (i.e., BERT, RoBERTa and GPT-2) demonstrate PST performs on par or better than previous sparsity methods.
  arXiv  Detail & Related papers  (2022-05-23T02:43:45Z)
• Powerpropagation: A sparsity inducing weight reparameterisation [65.85142037667065]
  We introduce Powerpropagation, a new weight- parameterisation for neural networks that leads to inherently sparse models. Models trained in this manner exhibit similar performance, but have a distribution with markedly higher density at zero, allowing more parameters to be pruned safely. Here, we combine Powerpropagation with a traditional weight-pruning technique as well as recent state-of-the-art sparse-to-sparse algorithms, showing superior performance on the ImageNet benchmark.
  arXiv  Detail & Related papers  (2021-10-01T10:03:57Z)
• Scaling Laws for Transfer [0.5432984841650929]
  We study scaling laws for transfer learning between distributions in an unsupervised, fine-tuning setting. We find that the effective data transferred is described well in the low data regime by a power-law of parameter count and fine-tuning dataset size.
  arXiv  Detail & Related papers  (2021-02-02T04:07:38Z)
• Understanding the Effects of Data Parallelism and Sparsity on Neural Network Training [126.49572353148262]
  We study two factors in neural network training: data parallelism and sparsity. Despite their promising benefits, understanding of their effects on neural network training remains elusive.
  arXiv  Detail & Related papers  (2020-03-25T10:49:22Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.