Skip to content
/ Consciences Public template

ConsciencesAI (TFBoost) Framework Repository (TensorFlow)

License

Notifications You must be signed in to change notification settings

Gabvaztor/Consciences

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Consciences || TensorFlow Boost Framework

This project is being moved to a new github repository. Also, the main project is being used in the url:

http://aiconscience.ddns.net/ai-models/

TensorFlow Boost's Repository

Windows CPU Windows GPU

Through project, we are creating a framework thanks to which you can read any kind of tag labeled data (like Kaggle problems, CSV and images); create train, validation and test set from them; choose the best machine learning algorithm for your data, and, besides, change the algorithm features.

Updating to Version 1.1:

  • New advanced and simplified structure
  • Using Tensorflow 2.0.0

Update Version 0.1:

  • Now it is possible to save models configurations by problem and load previous models configurations.
  • See json information by time in two different ways: "Information" and "Configuration".
  • See graphs progress during training. After each epoch it will be saved a graph.
  • You can decide if you want to save graphs after validation/test accuracies are surpass your limit.
  • You can reset the configuration making backups.
  • You can change dropout during training easily.
  • You can restore previous tensorflow models easily.
  • You have a method by problem: for each problem you can solve, you could created a method to process each input in a different way.
  • Yoy have a "Setting.json" file for each problem where you only have to put the paths where you want to process your problem.
  • You can see loss and accuracies in graphs and printed in the console.
  • You can easily change the epochs and batch sizes.
  • You have a CNN example treating a signal problem.
  • You have a Framework web to test models trained with TFBoost.

Next Version:

  • You will be able to do all this with a simple and beautiful user interface (in curse).
  • You will have an example of a LSTM project (in curse)

Future Versions:

  • In future, this project contains a graph visualization BEFORE TensorFlow generates his graphs.

All project use "Google Python Style Guide": https://google.github.io/styleguide/pyguide.html

TensorFlow Boost Web example (NEW):

http://aiconscience.ddns.net/ai-models/



TensorFlow Boost works as follows:



An example of 'information.json':



An example of a Accuracy Graph:



An example of a Loss Graph:



An example of code: Step by step structure (Python-Tensorflow Code)



"TensorFlow, the TensorFlow logo and any related marks are trademarks of Google Inc."