petered / artemis

Artemis aims to get rid of all the boring, bureaucratic coding (plotting, file management, organizing experiments, etc) involved in machine learning projects, so you can get to the good stuff quickly.

Artemis

Artemis is a collection of tools that make it easier to run experiments in Python. These include:

A simple framework for organizing your experiments and logging their results (text output and figures) so that they can be reviewed later and replicated easily.

e.g.

from artemis.experiments import experiment_function

@experiment_function  # Decorate your main function to turn it into an Experiment object
def multiply_3_numbers(a=1, b=2, c=3):
    answer = a*b*c
    print('{} x {} x {} = {}'.format(a, b, c, answer))
    return answer
    
record = multiply_3_numbers.run()  # Run experiment and save arguments, console output, and return value to disk
print(record.get_log())  # Pring console output of last run      
print(record.get_result())  # Print return value of last run
ex = multiply_3_numbers.add_variant(a=4, b=5)  # Make a new experiment with different paremters.
multiply_3_numbers.browse()  # Open a UI to browse through all experiments and results.

A dbplot function, for making live "debug" plots of numeric data on the fly.

e.g.

from artemis.plotting.db_plotting import dbplot
import numpy as np
for t in np.linspace(0, 10, 100):
    dbplot(np.sin(t), 'sin of the times')  # Detects data type and makes appropriate plot
    dbplot(np.sin(-4*t+np.sin(t/4.)*sum(xi**2 for xi in np.meshgrid(*[np.linspace(-20, 20, 200)]*2))), "Instaaaaall Arrrteeeemis")

(this can also be set up in the browser for remote live plotting)

Functions for easy download and loading of numerical data.

e.g.

from artemis.plotting.db_plotting import dbplot
from artemis.fileman.smart_io import smart_load
img = smart_load('https://cdn.britannica.com/s:700x450/54/13354-004-2F9AE1B2.jpg')  # Detects data type and loads into numpy array
dbplot(im, 'artemis', hang=True)

A system for downloading/caching files to a local directory, so the same code can work on different machines.

from artemis.fileman.file_getter import get_file
import os
local_path = get_file(url = 'https://cdn.britannica.com/s:700x450/54/13354-004-2F9AE1B2.jpg')  # Downloads first time, caches after 
print('Image "{}" has a size of {:.2g}kB'.format(local_path, os.path.getsize(local_path)/1000.))

For more examples of how to use artemis, read the Artemis Documentation

Installation

As of release 2.0.0 on November 13, 2017, Artemis now supports Python 3

To use artemis from within your project, use the following to install Artemis and its dependencies: (You probably want to do this in a virtualenv with the latest version of pip - run virtualenv venv; source venv/bin/activate; pip install --upgrade pip; to make one and enter it).

Option 1: Simple install (stable) :

pip install artemis-ml

Option 2: Install as source (master branch, do this if you want latest updates or you want to contribute to Artemis).

pip install -e git+https://github.com/petered/artemis.git#egg=artemis-ml

This will install it in (virtual env or system python root)/src/artemis. You can edit the code and submit pull requests to our git repo. To install with the optional remote plotting mode enabled, add the [remote_plotting] option, as in: pip install -e git+http://github.com/QUVA-Lab/artemis.git#egg=artemis[remote_plotting]

(Note, this doesn't work if you have Anaconda installed, as it does not work with the -e option). Use pip install artemis-ml in this case instead.

Verifying that it works

To verify that the plotting works, run:

python -m artemis.plotting.demo_dbplot

A bunch of plots should come up and start updating live.

Note: During installation, the settings file .artemisrc is created in your home directory. In it you can specify the plotting backend to use, and other settings.

Now that you have Artemis installed, see this Tutorial on how to use Artemis to organize your experiments.