Euler workshop 11.02.2021

Introduction

The HPC group organized a workshop for on-boarding the members of the Computer Vision Lab (CVL) from D-ITET to the Euler cluster.

Location

Online via Zoom

Time

10:00 - 15:00

Slides

Getting started with Euler
Using the batch system

Jobscript used in the ResNet live demo

Please note that you don't need the two lines for loading the modules, when you already executed the command (maybe in your .bashrc)

source /cluster/project/cvl/admin/cvl_settings

Jobscript:

#!/usr/bin/bash
#BSUB -n 128
#BSUB -W 4:00
#BSUB -R "rusage[mem=3500, ngpus_excl_p=8] span[hosts=1]"

# Load modules
source /cluster/apps/local/env2lmod.sh
module load gcc/6.3.0 python_gpu/3.8.5 eth_proxy

# TensorFlow codes with ResNet50 v1.5 model are in /cluster/project
PROJECT="/cluster/project/sis/cdss/jarunanp/sis/21_02_ml_benchmark"
TFMODELS=${PROJECT}/tf_official_models
export PYTHONPATH=${TFMODELS}:$PYTHONPATH

# The ImageNet2012 dataset is stored in /cluster/work
DATADIR="/cluster/work/sis/cdss/jarunanp/ml_benchmark/data"

# Train the model
python3 ${TFMODELS}/official/vision/image_classification/classifier_trainer.py \
--mode=train_and_eval \
--model_type=resnet \
--dataset=imagenet \
--model_dir=$PWD \
--data_dir=${DATADIR} \
--config_file=gpu.yaml