inputs must be a list of at least one Tensor with the same dtype and shape

[Lugangz]

It's a great job!!! When I run the cmd, python NIPS_figures.py --data artificial --save_dir layer1, an error happened like below, I wonder how to solve.

Traceback (most recent call last): File "/AdaptiveOptim/NIPS_figures.py", line 249, in network = LinearNetwork(D, 1, gpu_usage=gpu_usage, exp_dir=NAME_EXP) File "\AdaptiveOptim\adaopt\linear_network.py", line 28, in init super().init(n_layers=n_layers, name=name, **kwargs) File "\AdaptiveOptim\adaopt_loptim_network.py", line 22, in init self._construct() File "\AdaptiveOptim\adaopt_loptim_network.py", line 77, in _construct self._train = self._mk_training_step() File "\AdaptiveOptim\adaopt\linear_network.py", line 153, in _mk_training_step _reg = tf.add_n(tf.get_collection("regularisation")) File "C:\Anaconda3\envs\tensorflow-gpu\lib\site-packages\tensorflow\python\ops\math_ops.py", line 2222, in add_n raise ValueError("inputs must be a list of at least one Tensor with the " ValueError: inputs must be a list of at least one Tensor with the same dtype and shape

[scut-salmon]

same question. any update?

[scut-salmon]

There's Spelling mistake in _reg = tf.add_n(tf.get_collection("regularisation")) and it should be _reg = tf.add_n(tf.get_collection("regularization"))

[fengluyukaixin]

I met an error happened like below, I wonder how to solve. 'FactoNetwork' object has no attribute 'global_step', thank you very much.