examples

examples copied to clipboard

Remove deprecated model fit_generator

1

Remove: - `Model.fit_generator` method - and parameters `steps_per_epoch` and `validation_steps`

albertvillanova

Get class names from Dataset metadata

1

albertvillanova

Fix no module named 'sentencepiece' by install its package

6

There is an issue when importing TextClassifierDataLoader, with error message > ModuleNotFoundError: No module named 'sentencepiece'. Solved it by install sentencepiece.

andiwijayaid

Remove Dataset continue method

1

Remove: - `Dataset.continue` method - fit `steps_per_epoch` parameter - evaluate `steps` parameter

albertvillanova

In the Udacity course tf4dl: Changed 2 text boxes to match actual prediction

1

The markdown cells in l03c01_classifying_images_of_clothing.ipynb say that the model predicts the label 6, but when you run the code it correctly predicts the label 4. Maybe the order of the...

t4lz

[GSoC] TF 2.0 Custom ResNet Training

4

- Custom ResNet model for the TinyImageNet dataset without transfer learning. - Efficient model with `tf.keras.layers.SeparableConv2D` and `Resizing strategy` with ECLR to train with colab-like resource constrained environments. - TF...

Vishal-V

[GSoC] Autoencoder migration

2

- Migrated the autoencoder_runner.py and Autoencoder.py to Tensorflow 2.0 - The runner uses tf.data.Dataset pipeline to feed the batches and tf.keras.datasets to load the initial data and all necessary migrations...

Vishal-V

Delete download_models.sh

3

This file creates a /bin/bash^M: bad interpreter: No such file or directory error when building in Xcode. DOS uses a carriage return (^M) with a line feed therefore macOS (sharing...

nohren

Functional API supported. Model can now begin with tf.keras.Input() layer

3

Changed the way labels and inputs from `self._eval_signature` were read in `def train()` of `keras_model_head.py` such that it can support `tf.keras.Input()`.

sarthak-chakraborty

Output layer activation should be a softmax

1

In these example notebooks, the activation function on the output layer is not specified. According to the documentation, I believe they would be defaulting to 'None' (no function being applied...

kvenkman