MCPcopy
hub / github.com/MrNothing/AI-Blocks / Run

Function Run

Sources/scripts/classifier.py:24–57  ·  view source on GitHub ↗
(self)

Source from the content-addressed store, hash-verified

22save_path = ""
23
24def Run(self):
25
26 #defining the model...
27 self.X = tf.placeholder(tf.float32, shape=[None, self._input.input_size], name="x_input")
28 self.y = tf.placeholder(tf.float32, shape=[None, self._input.labels_size], name="y_labels")
29
30 _y = self._model.Run(self.X)
31 _loss = tf.reduce_mean(tf.pow(_y-self.y, 2))
32
33 correct_pred = tf.equal(tf.argmax(_y, 1), tf.argmax(self.y, 1))
34 _accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))
35
36 _solver = tf.train.AdamOptimizer(self.learning_rate).minimize(_loss)
37
38 instance = AIBlocks.InitModel(load_path=self.save_path)
39 Log("Model initialized!")
40
41 SetState(self.id, 0.001)
42
43 for it in range(self.training_iterations):
44 batch = self._input.getNextBatch()
45
46 X_batch = batch[0]
47 Y_batch = batch[1]
48
49 _, loss, acc = instance.Run([_solver, _loss, _accuracy], feed_dict={self.X: X_batch, self.y: Y_batch})
50
51 if it % self.display_step == 0:
52 SetState(self.id, it/self.training_iterations)
53 SendChartData(self.id, "Loss", loss, "#ff0000")
54 SendChartData(self.id, "Accuracy", acc)
55
56 AIBlocks.SaveModel(instance)
57 AIBlocks.CloseSession(instance)
58

Callers

nothing calls this directly

Calls 7

LogFunction · 0.85
SetStateFunction · 0.85
SendChartDataFunction · 0.85
InitModelMethod · 0.80
SaveModelMethod · 0.80
CloseSessionMethod · 0.80
RunMethod · 0.45

Tested by

no test coverage detected