Published September 27, 2022
| Version 1.0.1
Software
Open
DNN Mode Connectivity for TensorFlow
Authors/Creators
- 1. LMU Munich
Description
The content of this repository is based on the PyTorch implementation dnn-mode-connectivity by Timur Garipov and Pavel Izmailov, which is based on their paper Loss Surfaces, Mode Connectivity, and Fast Ensembling of DNNs.
This project was part of the Summer 2022 course Applied Deep Learning with TensorFlow and Pytorch at the LMU Munich, supervised by Dr. David Rügamer, Gregor Wiese, Tobias Weber und Daniel Dold.
Files
constantin-crailsheim/dnn-mode-connectivity-tf-1.0.1.zip
Files
(515.8 kB)
| Name | Size | Download all |
|---|---|---|
|
md5:fd5ac4836bb36ce2dbf52c5efb1d7baf
|
515.8 kB | Preview Download |
Additional details
Related works
- Is supplement to
- https://github.com/constantin-crailsheim/dnn-mode-connectivity-tf/tree/1.0.1 (URL)