Config Variants

This release lets you create "config variants" of your experiment, using experiment.add_config_variant(...). These let you specify how to construct arguments for your base experiment. Useful if your base experiment takes custom objects as arguments, and you want to define a set of variants specifying a particular way to construct these objects.

Artemis 2.0.0

Updates in this version

• Python 3 support
• Experiments can be defined as generators with "yield" keyword to save results periodically
• Can run experiments with Slurm
• Better UI display for narrow terminals
• The incredibly useful Duck data structure.

Updates to Experiment API

Some minor updates to make the experiment API more programmatically accessible.

Also some of the experiment UI function have been improved.

Release for Paper

Just a small patch on the last release to fix an error in determining whether a record existed for a given experiment.

Post NIPS Release

Merge pull request #48 from QUVA-Lab/ongoing_changes_2

Ongoing changes 2

0.1.0

• Adds a bunch of machine learning helpers to artemis.
• Fixes a bug to memoization.
• More options for dbplot

jewel-1: Merge pull request #14 from QUVA-Lab/recursive_experiments

Various add ons while doing spiking experiments