main_train does still nothing

[lBlitzdl]

Essentially this problem again: #38
But since a solution was never provided I reopen it.

I am using Google colab's free GPU to generate data and train models. I got data Generated, Data converted, but can't get Model trained.
The issue is whenever i run the training code, the script terminates at line 60 in Training/train.lua at the line:
M.network(inputs)
This function never returns, doesn't give any error or warning either.
Tried wirh different batch sizes, on CPU, on GPU

[airdine]

Hello,

Look at this issue : aikupoker#15

I used a fresh OS install to solve it.
By the way, if you use cuda > 9.2 I recommend to clone torch source from this repo :
https://github.com/nagadomi/distro.git

Hope it helps !

[lBlitzdl]

Thank you. Can you give me more information on how you did it?
Ubuntu Version?
Torch Version?
Lua Version?
Coda Version?

Thank you very much :)

[airdine]

The simplest way was to use this docker image :
nvidia/cuda:10.0-cudnn7-devel-ubuntu16.04

you can inspire yourself from : aikupoker repo

Good luck

[Sohaibb98]

for this, the host OS needs to be Ubuntu 16.04 or Ubuntu 18? on Ubuntu 18, it gives CMake error