An open API service indexing awesome lists of open source software.

https://github.com/stofte/tf-playground

Random tensorflow
https://github.com/stofte/tf-playground

jupyter-notebook machine-learning python3 tensorflow windows

Last synced: about 1 month ago
JSON representation

Random tensorflow

Awesome Lists containing this project

README

          

ML/Tensorflow playground (Windows/python36)
===============================

Recommended to use Anaconda for managing installation of components. Repo tested with Python 3. Currently [WSL does not support exposing GPU](https://github.com/Microsoft/WSL/issues/1788), limiting what use can be done usefully in the WSL Bash console on Windows. Windows/x64 specific Tensorflow [wheels](https://www.python.org/dev/peps/pep-0427/) for [Python can be found here](https://github.com/fo40225/tensorflow-windows-wheel).

MNIST dataset

- [Softmax regression](src/mnist_softmax.ipynb)
- [Convolutional neural net](src/mnist_convnn.ipynb)
- [Convnet using Keras](src/mnist_convnet_keras.ipynb)

[OpenAI Gym](https://gym.openai.com/)

- [CartPole](src/gym_cartpole.ipynb)

Software/library versions:

- Windows 10 (v1709)
- CUDA 9.1.85
- python 3.6.4
- anaconda 4.4.10
- jupyter 1.0
- tensorflow 1.7
- gym 0.10.5
- [GraphViz 2.38](https://graphviz.gitlab.io/_pages/Download/Download_windows.html) (Keras visualization)
- pyglet 1.2.4 (for `gym`)

Notes:

- Measure perf on Windows: `powershell -Command "Measure-Command { python script.py | Out-Default}"`