Lua docker: No value for $TERM


(Steve Braich) #1

Hello there. Sorry for the newbie question here.

I am using the the docker image from https://hub.docker.com/r/harvardnlp/opennmt. Apologies as this is my first time using docker. I did work thru a docker tutorial and created my own image, etc.

I assume when using the docker image you launch the first step (preprocessing) with the following command:

docker run harvardnlp/opennmt:8.0 th preprocess.lua -train_src data/src-train.txt -train_tgt data/tgt-train.txt -valid_src data/src-val.txt -valid_tgt data/tgt-val.txt -save_data data/demo

Aside from the docker prefix to this command, the rest is taken verbatim from the Lua Quickstart. I made sure the path is correct with corpora. However, I get the following error. It’s as if Torch is not installed in the Docker Image:

tput: No value for $TERM and no -T specified
/root/torch/install/bin/luajit: cannot open preprocess.lua: No such file or directory
stack traceback:
** [C]: in function ‘dofile’**
** /root/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:145: in main chunk**
** [C]: at 0x00406670**

Any ideas?

Update:
Just a quick follow-up. My understanding is that if I use the Docker Image, I don’t need to install Torch or anything else for that matter. Isn’t that correct? Thanks in advance.

System Specs:

  • Ubuntu 16.04
  • GTX 1070 GPU 8 GB
  • Intel i5 Quad Core
  • 16 GB RAM

2nd Update:
Sorry, I didn’t see these instructors for nvdia-docker. They were in the FAQs, which is an odd place to have them I think.
http://opennmt.net/FAQ/#what-type-of-computer-do-i-need-to-train-with