TensorFlow
Create your own model, customize it, train it and see some predictions right away.
This module is built on top of:
Python
Tensorflow
Keras
NLTK
Wordnet
Numpy
Flask
CORS
Alternatives we also excel in:
PyTorch
DeepPy
Django
Check out this dynamic tensorflow model! Customize your layers, optimizers, loss function and other metrics. Provide some trainig and testing data, and predict whatever you want!
Find below 2 examples already set for you to test. Play around with any variable including the predictions inputs and see results.
- - Example 1: We are training the model to learn about religion, sports and music; Give it any article and see the predictions.
- - Example 2: We are training the model to differentiate between water sports and winter sports; Provide any sentence and identify the type of the sport.
In general, training sets should hold thousands of records to be able to well train your model. Data is usually stored in a database, and improved over time. For the sake of keeping these demos simple, we are setting 10 records only, adding some synonyms on the texts and then applying the training. That's why the accuracy isn't high and the predictions are a bit inexact.