使用tensorflow执行如下最简单梯度下降的代码:
sess.run(tf.global_variables_initializer())
gradval =
sess.run(grads)
print(gradval)
256
if fetch is None:
257
raise
TypeError('Fetch argument %r has invalid type %r' %
(fetch,
259
elif isinstance(fetch, (list,
tuple)):
260
#
NOTE(touts): This is also the code path for namedtuples.
sess.run(tf.global_variables_initializer())
gradval =
sess.run(grads)
print(gradval)
import tensorflow as tf
w1 = tf.Variable([[1,2]])
w2 = tf.Variable([[3,4]])
y = tf.matmul(w1, [[9], [10]])
grads = tf.gradients(y, [w1])
with tf.Session() as sess:
报错信息最下边部分如下:
--> 258
type(fetch)))
TypeError: Fetch argument None has
invalid type
报错原因:该错误信息指的是grads操作没有结果返回,导致sess.run()的参数是NoneType。
进一步原因分析:
tensorflow gradients好像不支持 int型的Tensor
的gradients,如果把w1的设置成float类型的例如tf.float32就会执行成功。
修改为:
import tensorflow as tf
w1 = tf.Variable([[1.0,2.0]])
w2 = tf.Variable([[3.0,4.0]])
y = tf.matmul(w1, [[9.0], [10.0]])
grads = tf.gradients(y, [w1])
with tf.Session() as sess: