train.py keeps being killed

[Arlen22]

When running python train.py on docker tensorflow:1.0.1, the process keeps ending with "Killed".

[allenlu2009]

Same problem here. tensorflow: 1.2.1; Python 3.6.1; Anaconda 4.4.0. Run on google cloud platform using screen.

[andimarafioti]

I don't have this problem. Tried with tensorflow 1.3 (pip3 install...) and 1.3.1 (compiled from github) always with Python 3.4.2 and no GPU support. Also no Docker or Anaconda.

[rkuo2000]

got "Killed" with 2GB DRAM, after I tightly connect my 3 DRAM modules, python3 train.py keeps running with 6GB DRAM.
free -m shows it took ~2.3GB to run wavenet, so it should be safe to run with >3GB DRAM system.

[andimarafioti]

Mmm, the computer I tested this on has a huge amount of ram. It could be the case that you're just trying this without enough ram and that's why it's getting killed.

[philszalay]

same problem here. any ideas how to solve this?

[rkuo2000]

make sure you have enough memory (>3GB)

[philszalay]

I have expanded the memory in docker settings. Now everything works. Thank you!

[sakulh]

Hi, pleased could you tell me how?