Picture for Maxim Kodryan

Maxim Kodryan

Where Do Large Learning Rates Lead Us?

Add code
Oct 29, 2024
Viaarxiv icon

Large Learning Rates Improve Generalization: But How Large Are We Talking About?

Add code
Nov 19, 2023
Viaarxiv icon

Training Scale-Invariant Neural Networks on the Sphere Can Happen in Three Regimes

Add code
Sep 08, 2022
Figure 1 for Training Scale-Invariant Neural Networks on the Sphere Can Happen in Three Regimes
Figure 2 for Training Scale-Invariant Neural Networks on the Sphere Can Happen in Three Regimes
Figure 3 for Training Scale-Invariant Neural Networks on the Sphere Can Happen in Three Regimes
Figure 4 for Training Scale-Invariant Neural Networks on the Sphere Can Happen in Three Regimes
Viaarxiv icon

On the Periodic Behavior of Neural Network Training with Batch Normalization and Weight Decay

Add code
Jun 29, 2021
Figure 1 for On the Periodic Behavior of Neural Network Training with Batch Normalization and Weight Decay
Figure 2 for On the Periodic Behavior of Neural Network Training with Batch Normalization and Weight Decay
Figure 3 for On the Periodic Behavior of Neural Network Training with Batch Normalization and Weight Decay
Figure 4 for On the Periodic Behavior of Neural Network Training with Batch Normalization and Weight Decay
Viaarxiv icon

On Power Laws in Deep Ensembles

Add code
Jul 16, 2020
Figure 1 for On Power Laws in Deep Ensembles
Figure 2 for On Power Laws in Deep Ensembles
Figure 3 for On Power Laws in Deep Ensembles
Figure 4 for On Power Laws in Deep Ensembles
Viaarxiv icon

MARS: Masked Automatic Ranks Selection in Tensor Decompositions

Add code
Jun 18, 2020
Figure 1 for MARS: Masked Automatic Ranks Selection in Tensor Decompositions
Figure 2 for MARS: Masked Automatic Ranks Selection in Tensor Decompositions
Figure 3 for MARS: Masked Automatic Ranks Selection in Tensor Decompositions
Figure 4 for MARS: Masked Automatic Ranks Selection in Tensor Decompositions
Viaarxiv icon