rahulvigneswaran

import os

parser = argparse.ArgumentParser()
parser.add_argument("--gpu", default="0", type=str)



args = parser.parse_args()

rahulvigneswaran

Check this repo :

https://github.com/mseitzer/gpu-monitor

rahulvigneswaran

import os
import numpy as np
from scipy.stats import wasserstein_distance, energy_distance
from matplotlib import pyplot as plt 

epoch_name = "epoch_1_"
epoch = 1

a = os.getcwd()
root, dirs, files = next(os.walk(a))

rahulvigneswaran

import os
import numpy as np
from scipy.stats import wasserstein_distance, energy_distance
from matplotlib import pyplot as plt 

epoch_name = "epoch_1_"
epoch = 1

a = os.getcwd()
root, dirs, files = next(os.walk(a))

rahulvigneswaran

def print_nonzeros(model):
    nonzero = total = 0
    for name, p in model.named_parameters():
        tensor = p.data.cpu().numpy()
        nz_count = np.count_nonzero(tensor)
        total_params = np.prod(tensor.shape)
        nonzero += nz_count
        total += total_params
        print(f'{name:20} | nonzeros = {nz_count:7} / {total_params:7} ({100 * nz_count / total_params:6.2f}%) | total_pruned = {total_params - nz_count :7} | shape = {tensor.shape}')
    print(f'alive: {nonzero}, pruned : {total - nonzero}, total: {total}, Compression rate : {total/nonzero:10.2f}x  ({100 * (total-nonzero) / total:6.2f}% pruned)')

rahulvigneswaran

rahulvigneswaran

# First run the following things
echo "Downloading..."

wget -c http://dl.caffe.berkeleyvision.org/caffe_ilsvrc12.tar.gz

echo "Unzipping..."

tar -xf caffe_ilsvrc12.tar.gz && rm -f caffe_ilsvrc12.tar.gz

echo "Done."

rahulvigneswaran

#!/bin/bash
#
# script to download and extract ImageNet dataset
# ILSVRC2012_img_train.tar (about 138 GB)
# ILSVRC2012_img_val.tar (about 6.3 GB)
#
# make sure ILSVRC2012_img_train.tar & ILSVRC2012_img_val.tar in your current directory
#
#  https://github.com/facebook/fb.resnet.torch/blob/master/INSTALL.md
# 

rahulvigneswaran

Table of Contents

• CPU
• GPU
• Disk Space

CPU

• Dont use CPUs to train. This will slow experiments of others drastically. Almost always try to run your exps on GPUs excluding certain edge cases.

GPU

• Before you start a run on a GPU that someone else is already using, please do try and run first on an empty GPU, to check the GPU memory your experiment will occupy. If it exceeds the available memory (Use gpustat --watch to find the usage), PLEASE DONT RUN ON IT. This will crash the other person's runs.

rahulvigneswaran

Table of contents

• GPU
• CPU
• Storage
• General Training
  • Loss
  • Dataset/Dataloader
  • Misc