Meta-Principled Family of Hyperparameter Scaling Strategies

arXiv

Abstract

In this note, we first derive a one-parameter family of hyperparameter scaling strategies that interpolates between the neural-tangent scaling and mean-field/maximal-update scaling. We then calculate the scalings of dynamical observables – network outputs, neural tangent kernels, and differentials of neural tangent kernels – for wide and deep neural networks. These calculations in turn reveal a proper way to scale depth with width such that resultant large-scale models maintain their representation-learning ability. Finally, we observe that various infinite-width limits examined in the literature correspond to the distinct corners of the interconnected web spanned by effective theories for finite-width neural networks, with their training dynamics ranging from being weakly-coupled to being strongly-coupled.

Latest Publications

Presto: A Decade of SQL Analytics at Meta

Yutian James Sun, Tim Meehan, Rebecca Schlussel, Wenlei Xie, Masha Basmanova, Orri Erling, Andrii Rosa, Shixuan Fan, Rongrong Zhong, Arun Thirupathi, Nikhil Collooru, Ke Wang, Sameer Agarwal, Arjun Gupta, Dionysios Logothetis, Kostas Xirogiannopoulos, Bin Fan, Amit Dutta, Varun Gajjala, Rohit Jain, Ajay Palakuzhy, Prithvi Pandian, Sergey Pershin, Abhisek Saikia, Pranjal Shankhdhar, Neerad Somanchi, Swapnil Tailor, Jialiang Tan, Sreeni Viswanadha, Zac Wen, Deepak Majeti, Aditi Pandit, Biswapesh Chattopadhyay

SIGMOD - 2023