This is going to start out a being a little raw, but I’m going to start keeping my references here…

Learning Rates

batch norm

How do you know which parameters to choose for Keras (TensorFlow)?

This Example kernel, shown below, has some interesting properties.


#getting the 2D output:
output = model.get_layer("dense_3").output
extr = Model(model.input, output)



The following article is worth a look





Here’s another resource




Don’t ignore the power of Excel


Good Read…

http://www.scipy-lectures.org/

This is a good intro to python, for scientific computing:

cs231n.github.io/python-numpy-tutorial/

TODO

Look up Resnet

Python Examples for this book

RESOURCES

Deeplearning.net

This one seemed to have some cool code examples..

Numpy Stuff


# Options...left over plus any 1 
x[0]+leftOver
# will it increase the score?

# Come back to this
score = Score(x)

def lowest():
sum=100
tic=''
h=score[1]
for k in h.keys():
if h[k][0] < sum:
sum = h[k][0]
tic=k
ticN=np.fromstring(tic[1:-1], dtype=int, sep=' ')
return (tic,sum,ticN)

leftOver=total - np.sum(x,axis=0)


# What you have to work with
leftOver+lowest()[2]

def buildArrayFromScore():
a=np.array([])
for k in score[1].keys():
if a.size==0:
a=np.fromstring(k[1:-1], dtype=int, sep=' ')
else:
a=np.vstack((a,np.fromstring(k[1:-1], dtype=int, sep=' '))  ) 
return a


def ifFoundReplace(x,b,c):
if b == c:
return x
idx=np.where(np.all(x==b,axis=1))
if idx[0].size != 0:
x[idx]=np.array(c)
return x

score[1].keys()




Fast AI Course II

here

Django

Below are some quick notes on Django


mkdir ~/myproject2
cd myproject2
virtualenv -p python3 myprojectenv2
source myprojectenv2/bin/activate
pip install numpy
pip install pandas
pip install flask


Below is an example /etc/init.d/uwsgi script.

#!/bin/sh
description "uWSGI server"

start on runlevel [2345]
stop  on runlevel [!2345]
respawn
exec /home/webchirico/bin/bin/uwsgi -c /webproject/myproject/myproject.ini

[uwsgi]

module = wsgi

uid = www-data
gid = www-data
chdir = /webproject/myproject/
virtualenv = myprojectenv
wsgi-file = wsgi.py

master = true
processes = 5
enable-threads = true
thunder-lock = true

# For testing... user /spark39a
# http = 127.0.0.1:3031

socket = myproject.sock
chmod-socket = 666
vacuum = true

die-on-term = true

logto = /webproject/var/log/error.log
daemonize = /webproject/var/log/myproject.log


References

uwsgi-docs.readthedocs

stackoverflow

Steps for aide

aideinit -c /etc/aide/aide.conf -y -f

aide -c /etc/aide/aide.conf --check

aide -c /etc/aide/aide.conf --update

gsutil cp -r /var/lib/aide/aide.db gs://mchirico-aide/aipiggybot/

gsutil cp -r /etc/aide/aide.conf gs://mchirico-aide/aipiggybot/

Quick script

#!/bin/bash
export DATE=$(date "+%Y-%m-%d.%H:%M:%S")
aide -c /etc/aide/aide.conf -u > "aipiggybot.${DATE}"
gsutil cp "aipiggybot.${DATE}" gs://mchirico-aide/aipiggybot/
gsutil cp /etc/aide/aide.conf gs://mchirico-aide/aipiggybot/"aide.conf.${DATE}"
gsutil cp -r /var/lib/aide/aide.db gs://mchirico-aide/aipiggybot/
gsutil cp ./myaide.sh gs://mchirico-aide/aipiggybot/
gsutil ls gs://mchirico-aide/aipiggybot/
# We'll update
aideinit -c /etc/aide/aide.conf -y -f >/dev/null
cat "aipiggybot.${DATE}"
rm "aipiggybot.${DATE}"


  

Cool column matric idea.




 Another good list of lectures….

http://projects.iq.harvard.edu/stat110/youtube

Interesting blog on Neural Networks…