why train be killed? - Githubissues

CharlesShang / FastMaskRCNN

Mask RCNN in TensorFlow

Apache License 2.0

3.09k stars 1.1k forks source link

why train be killed? #172

Open landiaokafeiyan opened 6 years ago

landiaokafeiyan commented 6 years ago

classes
['background']
111111111begin iter 2222222222
iter 144: image-id:0254212, time:1.722(sec), regular_loss: 0.248765, total-loss 0.5882(0.0360, 0.5233, 0.000000, 0.0289, 0.0000), instances: 3, batch:(16|85, 0|11, 0|0)
labels
[]
classes
['background']

**Killed**

what is that mean?

VasanthBalakrishnan commented 6 years ago

I have seen this happen when the training process starts to take up most of the available memory and gets killed by linux. Can you give some details about the memory usage while training?