unexpected keyword argument 'top_k'

[Damon-wyg]

网络里并没有写过top_k，但是一直报top_k参数错误

F0623 09:15:01.814636 27241 PythonUtil.cpp:131] Check failed: (ret) != nullptr Current PYTHONPATH: ['/home/disk1/normandy/maybach/38769/workspace', '/home/disk1/normandy/maybach/38769/workspace/python27-gcc482/lib/python2.7/site-packages/setuptools-18.2-py2.7.egg', '/home/disk1/normandy/maybach/38769/workspace/paddle', '/home/disk1/normandy/maybach/38769/workspace', '/home/disk1/normandy/maybach/38769/workspace/thirdparty/thirdparty', '/home/disk1/normandy/maybach/38769/workspace/python27-gcc482/lib/python27.zip', '/home/disk1/normandy/maybach/38769/workspace/python27-gcc482/lib/python2.7', '/home/disk1/normandy/maybach/38769/workspace/python27-gcc482/lib/python2.7/plat-linux2', '/home/disk1/normandy/maybach/38769/workspace/python27-gcc482/lib/python2.7/lib-tk', '/home/disk1/normandy/maybach/38769/workspace/python27-gcc482/lib/python2.7/lib-old', '/home/disk1/normandy/maybach/38769/workspace/python27-gcc482/lib/python2.7/lib-dynload', '/home/disk1/normandy/maybach/38769/workspace/python27-gcc482/lib/python2.7/site-packages']
Python Error: <type 'exceptions.TypeError'> : precision_recall_evaluator() got an unexpected keyword argument 'top_k'
Python Callstack: 
            /home/disk1/normandy/maybach/38769/workspace/python27-gcc482/lib/python2.7/site-packages/paddle/trainer/config_parser.py : 3607
            /home/disk1/normandy/maybach/38769/workspace/python27-gcc482/lib/python2.7/site-packages/paddle/trainer/config_parser.py : 3600
            conf/trainer_config.conf : 99
            /home/disk1/normandy/maybach/38769/workspace/python27-gcc482/lib/python2.7/site-packages/paddle/trainer_config_helpers/default_decorators.py : 53
            /home/disk1/normandy/maybach/38769/workspace/python27-gcc482/lib/python2.7/site-packages/paddle/trainer_config_helpers/layers.py : 329
            /home/disk1/normandy/maybach/38769/workspace/python27-gcc482/lib/python2.7/site-packages/paddle/trainer_config_helpers/layers.py : 3850
            /home/disk1/normandy/maybach/38769/workspace/python27-gcc482/lib/python2.7/site-packages/paddle/trainer_config_helpers/layers.py : 3844
            /home/disk1/normandy/maybach/38769/workspace/python27-gcc482/lib/python2.7/site-packages/paddle/trainer_config_helpers/default_decorators.py : 53
Call Object failed.
Path	URL	2.39KB	

[qingqing01]

You use classification_cost as follows:

cls = classification_cost(input=out_layer, label=label, evaluator=[precision_recall_evaluator])

As @lcy-seso said, this problem had been fixed in PR. But the paddle version in the cluster has not been updated yet.

But you can use another method:

cost= cross_entropy(out_layer, label)
eval=precision_recall_evaluator(out_layer, label)