Loss Landscapes are All You Need: Neural Network Generalization Can Be Explained Without the Implicit Bias of Gradient Descent

Ping-yeh Chiang, Renkun Ni, David Yu Miller, Arpit Bansal, Jonas Geiping +2 more
2/12/2026

No official code found. Best implementation: Ping-C/optimizer (39 stars, last commit 2023-03-02).

Semantic Scholar

Code Implementations(1)

Ping-C/optimizerCommunity67%

This repository includes code to reproduce the tables in "Loss Landscapes are All You Need: Neural Network Generalization Can Be Explained Without the Implicit Bias of Gradient Descent"

395Shell, PythonFeb 28, 20233 years ago

Cite this paper

@article{chiang2026loss,
  title  = {Loss Landscapes are All You Need: Neural Network Generalization Can Be Explained Without the Implicit Bias of Gradient Descent},
  author = {Ping-yeh Chiang and Renkun Ni and David Yu Miller and Arpit Bansal and Jonas Geiping and Micah Goldblum and T. Goldstein},
  year   = {2026},
  url    = {https://api.semanticscholar.org/CorpusID:e4f0ab1408504c3873a849d341ec63fbca899534},
  journal = {ICLR 2023 2023}
}

Discussion