random.seed

[11sc]

tf.reset_default_graph()
np.random.seed(10)
rand_x = np.random.randn(1500)/50
np.random.seed(8)
rand_y = np.random.randn(1500)/50

Why there are two seed in the data initialization part in toydata exp.?

[bisno]

Hi, the seed here helps to generate random toy data. As long as the total training data construct the pattern region (the grey area in the readme ), it will not affect the final outcome.