https://github.com/tensorpack/benchmarks
Use TensorFlow efficiently
https://github.com/tensorpack/benchmarks
Last synced: about 1 year ago
JSON representation
Use TensorFlow efficiently
- Host: GitHub
- URL: https://github.com/tensorpack/benchmarks
- Owner: tensorpack
- License: unlicense
- Created: 2017-09-26T22:19:47.000Z (over 8 years ago)
- Default Branch: master
- Last Pushed: 2021-05-07T04:02:45.000Z (about 5 years ago)
- Last Synced: 2025-03-23T20:06:02.050Z (about 1 year ago)
- Language: Python
- Homepage:
- Size: 188 KB
- Stars: 95
- Watchers: 8
- Forks: 32
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# tensorpack benchmarks
We use TensorFlow in the efficient way. Tensorpack is:
* [As fast as tensorflow/benchmarks in multi-GPU ResNet training](ResNet-MultiGPU/)
* [__1.2x~5x__ faster than Keras & tflearn in common CNNs](other-wrappers/)
* [Able to reproduce "ImageNet in one hour" with 256 GPUs](ResNet-Horovod/)
* [Able to train Cifar10 to 94% accuracy within __a minute__](Cifar10-fast)
* [__5x__ faster than matterport/Mask_RCNN](MaskRCNN/)
* [2.8x faster than DCGAN-tensorflow](DCGAN/)
All above claims can be reproduced with the corresponding code.