The method tensorflow.nn.xw_plus_b() exists but is not documented in the online API documentation.
I think this is intentional: it's better to use tf.matmul(x, w) + b
directly. We are working to lock down our symbol exports so that only documented symbols are exported, but we're a bit lax at the moment.
tf.nn.xw_plus_b()
has option to name the operation whiletf.matmul(x, w) + b
has not. Is there any way around that?
@dParadiz you can just use tf.add(tf.matmul(x, w), b, name='
Most helpful comment
I think this is intentional: it's better to use
tf.matmul(x, w) + b
directly. We are working to lock down our symbol exports so that only documented symbols are exported, but we're a bit lax at the moment.