Mercurial > repos > bgruening > run_jupyter_job

comparison test-data/tf-script.py @ 0:f4619200cb0a draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/jupyter_job commit f945b1bff5008ba01da31c7de64e5326579394d6"
author bgruening
date Sat, 11 Dec 2021 17:56:38 +0000
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:f4619200cb0a
1 import numpy as np
2 import tensorflow as tf
3
4 (mnist_images, mnist_labels), _ = tf.keras.datasets.mnist.load_data()
5 mnist_images, mnist_labels = mnist_images[:128], mnist_labels[:128]
6 dataset = tf.data.Dataset.from_tensor_slices((tf.cast(mnist_images[..., tf.newaxis] / 255, tf.float32), tf.cast(mnist_labels, tf.int64)))
7 dataset = dataset.shuffle(1000).batch(32)
8
9 tot_loss = []
10 epochs = 1
11
12 mnist_model = tf.keras.Sequential([
13 tf.keras.layers.Conv2D(16, [3, 3], activation='relu'),
14 tf.keras.layers.Conv2D(16, [3, 3], activation='relu'),
15 tf.keras.layers.GlobalAveragePooling2D(),
16 tf.keras.layers.Dense(10)
17 ])
18
19 optimizer = tf.keras.optimizers.Adam()
20 loss_object = tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True)
21
22 for epoch in range(epochs):
23 loss_history = []
24 for (batch, (images, labels)) in enumerate(dataset):
25 with tf.GradientTape() as tape:
26 logits = mnist_model(images, training=True)
27 loss_value = loss_object(labels, logits)
28 loss_history.append(loss_value.numpy().mean())
29 grads = tape.gradient(loss_value, mnist_model.trainable_variables)
30 optimizer.apply_gradients(zip(grads, mnist_model.trainable_variables))
31 tot_loss.append(np.mean(loss_history))