yhlleo / tensorflow.cifar10 Goto Github PK

https://blog.csdn.net/xiao_lxl/article/details/70622209

apply_gradient_op = opt.apply_gradients(grads, global_step=global_step)
Traceback (most recent call last):
File "", line 1, in
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/training/optimizer.py", line 462, in apply_gradients
apply_updates = state_ops.assign_add(global_step, 1, name=name).op
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/ops/state_ops.py", line 240, in assign_add
return ref.assign_add(value, name=name)
AttributeError: 'Tensor' object has no attribute 'assign_add'

打扰请问
cifar10_input_test.py 和 compat.py 的作用是什么呢

运行train_op = train(loss, global_step)，我是train_op = train(loss, 1000)，然后报错，我一行一行的测试，在
apply_gradient_op = opt.apply_gradients(grads, global_step=1000.0)这里错了，

Traceback (most recent call last):
File "", line 1, in
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/training/optimizer.py", line 462, in apply_gradients
apply_updates = state_ops.assign_add(global_step, 1, name=name).op
File "/usr/local/lib/python2.7/dist-packages/tensorflow/python/ops/state_ops.py", line 237, in assign_add
if ref.dtype._is_ref_dtype:
AttributeError: 'float' object has no attribute 'dtype'

请问为什么？