https://github.com/tonyz0x0/ts-cnn-benchmark

The CNN benchmark test on Tensorflow
https://github.com/tonyz0x0/ts-cnn-benchmark

python tensorflow

Last synced: 2 months ago
JSON representation

The CNN benchmark test on Tensorflow

• Host: GitHub
• URL: https://github.com/tonyz0x0/ts-cnn-benchmark
• Owner: tonyz0x0
• License: mit
• Created: 2019-04-12T13:37:03.000Z (about 7 years ago)
• Default Branch: master
• Last Pushed: 2019-06-13T03:05:21.000Z (about 7 years ago)
• Last Synced: 2025-02-02T08:27:43.651Z (over 1 year ago)
• Topics: python, tensorflow
• Language: Python
• Homepage:
• Size: 113 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# cnn

- Varied the size of training data

Change the batch size can achieve the goal.

- Varied the dataset

We use **cifar** and **svhn** these two datasets.

- Varied the version of TensorFlow application(CPU version or GPU version)

To use CPU and GPU can explicitly be configured by the options variable.

- Varied the number of CPU or GPU

The number of CPUs and GPUs can explicitly be configured by the options variable

https://ryannng.github.io/2016/12/20/Street-View-House-Numbers-Recognition-Using-ConvNets/

https://stackoverflow.com/questions/37660312/how-to-run-tensorflow-on-cpu

https://stackoverflow.com/questions/41233635/meaning-of-inter-op-parallelism-threads-and-intra-op-parallelism-threads/41233901#41233901

Profile Module Usage: https://zhuanlan.zhihu.com/p/40156908

## How to run

```bash

# Run tensorboard

$ tensorboard --logdir ~/workplace/cnn/logs/cifar10_eval

# Run train

$  python src/cifar10_train.py

# Run eval

$ python src/cifar10_eval.py

```