Grokking in Linear Estimators -- A Solvable Model that Groks without Understanding
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2023
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913222740672512 |
|---|---|
| author | Levi, Noam Beck, Alon Bar-Sinai, Yohai |
| author_facet | Levi, Noam Beck, Alon Bar-Sinai, Yohai |
| contents | Grokking is the intriguing phenomenon where a model learns to generalize long after it has fit the training data. We show both analytically and numerically that grokking can surprisingly occur in linear networks performing linear tasks in a simple teacher-student setup with Gaussian inputs. In this setting, the full training dynamics is derived in terms of the training and generalization data covariance matrix. We present exact predictions on how the grokking time depends on input and output dimensionality, train sample size, regularization, and network initialization. We demonstrate that the sharp increase in generalization accuracy may not imply a transition from "memorization" to "understanding", but can simply be an artifact of the accuracy measure. We provide empirical verification for our calculations, along with preliminary results indicating that some predictions also hold for deeper networks, with non-linear activations. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2310_16441 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Grokking in Linear Estimators -- A Solvable Model that Groks without Understanding Levi, Noam Beck, Alon Bar-Sinai, Yohai Machine Learning Disordered Systems and Neural Networks Mathematical Physics Grokking is the intriguing phenomenon where a model learns to generalize long after it has fit the training data. We show both analytically and numerically that grokking can surprisingly occur in linear networks performing linear tasks in a simple teacher-student setup with Gaussian inputs. In this setting, the full training dynamics is derived in terms of the training and generalization data covariance matrix. We present exact predictions on how the grokking time depends on input and output dimensionality, train sample size, regularization, and network initialization. We demonstrate that the sharp increase in generalization accuracy may not imply a transition from "memorization" to "understanding", but can simply be an artifact of the accuracy measure. We provide empirical verification for our calculations, along with preliminary results indicating that some predictions also hold for deeper networks, with non-linear activations. |
| title | Grokking in Linear Estimators -- A Solvable Model that Groks without Understanding |
| topic | Machine Learning Disordered Systems and Neural Networks Mathematical Physics |
| url | https://arxiv.org/abs/2310.16441 |