NVIDIA Deep Learning Examples
NVIDIA Deep Learning Examples
ResNet50 v1.5 for TensorFlow1
Resource
NVIDIA Deep Learning Examples
NVIDIA Deep Learning Examples
ResNet50 v1.5 for TensorFlow1

With modified architecture and initialization this ResNet50 version gives ~0.5% better accuracy than original.

Changelog

  1. March, 2019
  • Initial release
  1. May, 2019
  • Added DALI support
  • Added scripts for DGX-2
  • Added benchmark results for DGX-2 and XLA-enabled DGX-1 and DGX-2.
  1. July, 2019
  • Added Cosine learning rate schedule
  1. August, 2019
  • Added mixup regularization
  • Added T4 benchmarks
  • Improved inference capabilities
  • Added SavedModel export
  1. January, 2020
  • Removed manual checks for dataset paths to facilitate cloud storage solutions
  • Move to a new logging solution
  • Bump base docker image version
  1. March, 2020
  • Code cleanup and refactor
  • Improved training process
  1. June, 2020
  • Added benchmark results for DGX-A100
  • Updated benchmark results for DGX-1, DGX-2 and T4
  • Updated base docker image version
  1. August 2020
  • Updated command line argument names
  • Added support for syntetic dataset with different image size
  1. January, 2022
  • Added barrier at the end of multiprocess run

Known issues

Performance without XLA enabled is low due to BN + ReLU fusion bug.

NVIDIA uses cookies to improve your experience on our web site. We and our third-party partners also use cookies and other tools to collect and record information you provide as well as information about your interactions with our websites for performance improvement, analytics, and to assist in marketing efforts. By clicking "Accept All", you consent to our use of cookies and other tools as described in our Cookie Policy. You can manage your cookie settings by clicking on "Manage Settings." By continuing to use this site or by clicking one of the buttons below, you agree to our Terms of Service (which contains important waivers). Please see our Privacy Policy for more information on our privacy practices.