shartoo / luna16_multi_size_3dcnn

An implement of paper "Multi-level Contextual 3D CNNs for False Positive Reduction in Pulmonary Nodule Detection"
111 stars 45 forks source link

new_model.py #19

Open chenyafang opened 5 years ago

chenyafang commented 5 years ago

new_model.py和model.py的区别是什么,是不是运行new_model.py就是训练模型了,就不用再运行model.py了。谢谢!

shartoo commented 5 years ago

model.py的代码好像是有点问题的,建议使用new_model.py

chenyafang commented 5 years ago

请问python new_model.py是训练模型,运行new_model.py需要很大的内存吗,我的怎么总是出现ResourceExhaustedError (see above for traceback): OOM when allocating tensor with shape[1000,64,6,20,20] [[Node: arch-1/MaxPool3D = MaxPool3DT=DT_FLOAT, data_format="NDHWC", ksize=[1, 1, 1, 1, 1], padding="SAME", strides=[1, 1, 1, 1, 1], _device="/job:localhost/replica:0/task:0/device:GPU:0"]] 是内存不够吗?但是我已经把batch_size调小了好多,还是出现这种错误。请问该怎样解决,谢谢!

shartoo commented 5 years ago

batch_size 建议为32或者16吧。

chenyafang commented 5 years ago

好的,谢谢。

chenyafang commented 5 years ago

每次训练的pos_mun+neg_num=320,比如,第一轮训练pos_num=2,neg_num=318,为什么总和是320呢?谢谢!

shartoo commented 5 years ago

好像没有这样的指定啊?