개요

NSML에서 샘플코드로 submit 확인하는 방법을 간단히 정리한다.

code clone

$ git clone https://github.com/naver/ai-hackathon-2018
$ cd ai-hackathon-2018/missions/examples/kin/example/

ls

$ nsml dataset ls

run

$ nsml run -d kin_phase1

ps

(tf27_r17_dev)lks21c@lks21c-desktop-ubuntu:example (master)$ nsml ps -a
Name                   Created         Args     Status    Summary                                    Description
---------------------  --------------  -------  --------  -----------------------------------------  -------------
lks21c/kin_phase1/149  seconds ago     main.py  Running   epoch=7, epoch_total=10, train/loss=0.623
lks21c/kin_phase1/148  25 minutes ago  run.py   Exited

ls

nsml model ls lks21c/kin_phase1/149
Checkpoint    Last Modified    Elapsed    Summary
------------  ---------------  ---------  --------------------------------------------------------------
0             a minute ago     0.328      epoch=0, train/loss=0.7778596252202987, step=0, epoch_total=10
1             a minute ago     1.555      epoch=1, train/loss=0.69048513174057, step=1, epoch_total=10
2             a minute ago     1.538      epoch=2, train/loss=0.6461338549852371, step=2, epoch_total=10
3             a minute ago     1.570      epoch=3, train/loss=0.6403789028525353, step=3, epoch_total=10
4             a minute ago     1.556      epoch=4, train/loss=0.637733145058155, step=4, epoch_total=10
5             a minute ago     1.549      epoch=5, train/loss=0.6321869477629661, step=5, epoch_total=10
6             a minute ago     1.607      epoch=6, train/loss=0.6279186278581619, step=6, epoch_total=10
7             a minute ago     7.907      epoch=7, train/loss=0.6234762459993363, step=7, epoch_total=10
8             seconds ago      1.622      epoch=8, train/loss=0.6195324093103409, step=8, epoch_total=10
9             seconds ago      3.907      epoch=9, train/loss=0.6160384342074394, step=9, epoch_total=10

submit

$ nsml submit nsml-admin/kin_phase1/3 9
$ nsml submit nsml-admin/kin_phase1/3 모델명 # 케라스로 학습시켰을때 이걸로 해결

#

참고자료