Created by: sonofaeolus

Traceback (most recent call last): 
File "policy_gan.py", line 74, in <module> train()
File "policy_gan.py", line 52, in train opt.minimize(loss = dg_loss, parameter_list = g_params)
File "/home/work/wangxiaoyu/fpython/lib/python2.7/site-packages/paddle/fluid/optimizer.py", line 259, in minimize [error_clip_callback])
File "/home/work/wangxiaoyu/fpython/lib/python2.7/site-packages/paddle/fluid/backward.py", line 590, in append_backward _append_backward_vars_(root_block, fwd_op_num, grad_to_var, grad_info_map)
File "/home/work/wangxiaoyu/fpython/lib/python2.7/site-packages/paddle/fluid/backward.py", line 426, in _append_backward_vars_ op_desc.infer_shape(block.desc) 
paddle.fluid.core.EnforceNotMet: Input(X) of LoDTensorToArrayOp should not be null. at [/paddle/paddle/fluid/operators/lod_tensor_to_array_op.cc:183]