iamaaditya

# memory footprint support libraries/code
!ln -sf /opt/bin/nvidia-smi /usr/bin/nvidia-smi
!pip install gputil
!pip install psutil
!pip install humanize
import psutil
import humanize
import os
import GPUtil as GPU
GPUs = GPU.getGPUs()

iamaaditya

FILENAME = paper
PS_FILE = $(FILENAME).ps
PDF_FILE = $(FILENAME).pdf

# PAPERSIZE = a4
PAPERSIZE = letter

LATEX_FILES = *.dvi *.log *.toc *.tof *.aux *.blg *.lof *.lot *.bbl
CLEAN_FILES = $(LATEX_FILES) *.bak core $(PS_FILE) $(PDF_FILE)
COMPRESS_FILES = *.tex *.bib *.sty *.eps *.ps *.fig *.m *.txt *.pgm *.bst *.cls

iamaaditya

%matplotlib inline

import matplotlib
import numpy as np
import pandas as pd
from glob import glob
import matplotlib.pyplot as plt
plt.style.use('ggplot')
import seaborn as sns; sns.set()

iamaaditya

""" This program takes a file, and splits it into given percentage by line number, but uses
randomization to select the files 
USAGE: python randomize_split.py <file_name> <split_percentage>
@author: aaditya prakash"""

from __future__ import division
import sys
import random

iamaaditya

def get_variable_with_decay(name, shape, initializer, lambda):
  if initializer is None:
    initializer = tf.truncated_normal_initializer(stddev=5e-2, dtype=tf.float32)
  var = tf.get_variable(name, shape, initializer=initializer, dtype=tf.float32)
    
  if lambda is not None:
    weight_decay = tf.multiply(tf.nn.l2_loss(var), lambda, name=name+'_weight_decay')
    tf.add_to_collection('losses', weight_decay)
    
  return var

iamaaditya

import os,sys
import numpy as np
import tensorflow as tf
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
from util.data import get_data_NCHW_normalized as get_data
from model import n_way
from random import shuffle
from time import time

def chunker(seq, size):

iamaaditya

import numpy as np
from keras.layers import Dropout
from keras import applications
from keras.layers import Dense, GlobalAveragePooling2D, merge, Input
from keras.models import Model

max_words = 10000
epochs = 50
batch_size = 32

iamaaditya

# XLA compilation controlled by "compile_ops" option
# compile_ops=False: 4.39 sec
# compile_ops=True: 0.90 sec


import os
os.environ['CUDA_VISIBLE_DEVICES']=''
import tensorflow as tf

iamaaditya

# Code snippet to print various ML related metrics given the y_labels and probabilities of each label (output of softmax)
# Aaditya Prakash

from sklearn.metrics import f1_score, roc_auc_score, precision_score, recall_score, accuracy_score, average_precision_score, precision_recall_curve, hamming_loss

def print(y_labels, probs):

    threshold = 0.5
    macro_auc                  = roc_auc_score(y_labels, probs, average            = 'macro')
    micro_auc                  = roc_auc_score(y_labels, probs, average            = 'micro')

iamaaditya

Keybase proof

I hereby claim:

• I am iamaaditya on github.
• I am iamaaditya (https://keybase.io/iamaaditya) on keybase.
• I have a public key whose fingerprint is 00B9 886D 6185 B18A 290A 02B8 2C21 2B82 7EC6 0287

To claim this, I am signing this object: