hiwonjoon

<<< Depreciated >>>

1. Conda Install

wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
chmod +x Miniconda3-latest-Linux-x86_64.sh
bash Miniconda3-latest-Linux-x86_64.sh
<answer appropriately>
<recommend to setup conda on /scratch/cluster/<username>. Since miniconda can become really big!>

hiwonjoon

Updated 4/11/2018

Here's my experience of installing the NVIDIA CUDA kit 9.0 on a fresh install of Ubuntu Desktop 16.04.4 LTS.

Table of Contents

• Install NVIDIA Graphics Driver via apt-get
• Install CUDA
• Install cuDNN

Table of contents generated with markdown-toc

hiwonjoon

import numpy as np
from matplotlib import pyplot as plt
from matplotlib import animation, rc
from IPython.core.display import display, HTML

%matplotlib inline
%load_ext autoreload
%autoreload 2

def display_frames_as_gif(frames):

hiwonjoon

#!/bin/bash

set -e

TMUX_VERSION=2.4
LIBEVENT_VERSION=2.0.22-stable
NCURSES_VERSION=6.0

mkdir -p $HOME/.local
mkdir -p $HOME/src

hiwonjoon

import better_exceptions
import os
import numpy as np
from tqdm import tqdm

from keras.models import model_from_json
import c3d_model
import sys
import keras.backend as K
dim_ordering = K.image_dim_ordering()

hiwonjoon

# This is shorthened version of blog post 
#  http://ksopyla.com/2017/02/tensorflow-gpu-virtualenv-python3/

# update packages
sudo apt-get update
sudo apt-get upgrade

#Add the ppa repo for NVIDIA graphics driver
sudo add-apt-repository ppa:graphics-drivers/ppa
sudo apt-get update

hiwonjoon

If you want to know how to reproduce this results, come to visit my github repo! 🐶

hiwonjoon

import numpy as np
import sys
import random
import tensorflow as tf
import itertools
slim = tf.contrib.slim

BATCH_SIZE = 32
TRAIN_ITER = 1000
EARLY_TRAIN_TERMINATE_CRITERIA = 1e-3;

hiwonjoon

import numpy as np
import sys
import random
import tensorflow as tf
import itertools
slim = tf.contrib.slim

BATCH_SIZE = 32
TRAIN_ITER = 1000
EARLY_TRAIN_TERMINATE_CRITERIA = 1e-4;

hiwonjoon

import subprocess
import datetime
import numpy as np

THREAD_NUM=4

def get_video_info(fileloc) :
    command = ['ffprobe',
               '-v', 'fatal',
               '-show_entries', 'stream=width,height,r_frame_rate,duration',