Скачать или смотреть 30: Transfer Learning | TensorFlow | Tutorial

The video discusses in TensorFlow: Transfer learning or Retraining a Classifier
[Note: Using pre-built model from TensorFlow Hub]

00:00:00 - Overview
00:03:20 - Import libraries
00:06:08 - Select TF2 saved model from TensorFlow Hub: "efficientnetv2-xl-21k"
00:08:12 - Create a model_handle
00:09:40 - Set IMAGE_SIZE, BATCH_SIZE
00:10:30 - Download 'flower_photos' dataset: tf.keras.utils.get_file()
00:13:00 - Build dataset: Create a function: tf.keras.preprocessing.image_dataset_from_directory()
00:15:38 - Build dataset: train_ds
00:17:09 - Build dataset: normalization_layer: tf.keras.layers.Rescaling(1./255)
00:17:57 - Build dataset: preprocessing: tf.keras.Sequential([normalization_layer])
00:18:35 - Build dataset: data augmentation: tf.keras.layers.RandomRotation()
00:19:26 - Build dataset: data augmentation: tf.keras.layers.RandomTranslation()
00:20:30 - Build dataset: data augmentation: tf.keras.layers.RandomZoom()
00:21:00 - Build dataset: data augmentation: tf.keras.layers.RandomFlip()
00:21:33 - Build dataset: apply augmentation to train_ds
00:22:22 - Build dataset: validation
00:24:08 - Define the model: tf.keras.layers.Sequential()
00:25:41 - Define the model: hub.KerasLayers()
00:28:04 - Build model: model.build(), model.summary()
00:29:04 - Train model: model.compile()
00:30:30 - Fit model: model.fit()
00:32:52 - Take a look at CPU, System RAM and GPU RAM resources used!
00:34:34 - Fit model: edit steps per epoch and validation steps
00:36:16 - Plot: accuracy and loss for train and validation
00:39:13 - Predict: get data
00:41:03 - Predict: model.predict()
00:42:50 - Save model: tf.saved_model.save()
00:45:50 - Deployment to TensorFlow Lite: * * * No output * * *
00:47:30 - Optimization: representative dataset
00:47:41 - Converter: tf.lite.TFLiteConverter.from_saved_model()
00:47:53 - TF Lite model: converter.convert() * * * Code crashed here * * *
00:48:10 - Write .tfile to file: f.write()
00:48:17 - Interpreter: tf.lite.Interpreter()
00:48:45 - Evaluate tflite model * * * No output * * *
00:49:16 - Ending notes


----------------
TensorFlow Guide
----------------
https://www.tensorflow.org/hub/tutori...