usage memo about google cloud platform / deep learning vm

memo.org

get gpu

1. “IAM and settings”
2. “assignment”
3. check “GPUs(all regions)” and click “assignment settings”
4. input your phone number
5. input some information in English
6. send request

deploy “Deep Learning VM”

enable http/https connection

enable pre-install cuda/tensorflow

ssh connection via web browser

create save storage

1. storage
2. browser
3. create a packet

run these code

  gcloud auth login
  python /usr/share/tensorflow/tensorflow/examples/how_tos/reading_data/convert_to_records.py 
--directory=./data
  gunzip ./data/*.gz
  export STORAGE_BUCKET=gs://[your storage name]
  gsutil cp -r ./data ${STORAGE_BUCKET}

and delete storage for saving money…lol

see. https://cloud.google.com/ml-engine/docs/tensorflow/getting-started-training-prediction

in **(your vm)**

wget https://github.com/GoogleCloudPlatform/cloudml-samples/archive/master.zip
unzip master.zip
cd cloudml-samples-master/census/estimator
mkdir data
gsutil -m cp gs://cloud-samples-data/ml-engine/census/data/* data/
TRAIN_DATA=$(pwd)/data/adult.data.csv
EVAL_DATA=$(pwd)/data/adult.test.csv
pip install --user -r ../requirements.txt
MODEL_DIR=output
gcloud ml-engine local train \
    --module-name trainer.task \
    --package-path trainer/ \
    --job-dir $MODEL_DIR \
    -- \
    --train-files $TRAIN_DATA \
    --eval-files $EVAL_DATA \
    --train-steps 1000 \
    --eval-steps 100
tensorboard --logdir=$MODEL_DIR --port=6006 & 

for opening tensorboard, we should create ssh tonnel

in **(cloudshell)**

gcloud compute ssh $INSTANCE_NAME -- -L 6006:localhost:6006

open tensorboard

click “preview using web browser” with port 6006