Open cometta opened 4 years ago
In your Attention class, self.X shape is [1 5 1], after executing below x_mean = tf.reduce_mean(self.X, axis = 2) the shape becomes [1 5], if mean is apply on axis=2 , the values will be the same, and only reduce [1 5 1] to [1 5] . Is this correct?
mean over axis 2 on [1 5 1] will be reduced to [1 5] yes
In your Attention class, self.X shape is [1 5 1], after executing below x_mean = tf.reduce_mean(self.X, axis = 2) the shape becomes [1 5], if mean is apply on axis=2 , the values will be the same, and only reduce [1 5 1] to [1 5] . Is this correct?