Skip to content
forked from martno/exprec

Exprec records your execution runs so you can compare different experiments and more easily reproduce results

License

Notifications You must be signed in to change notification settings

mehdi-shiba/exprec

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

41 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Exprec

Exprec records your execution runs so you can compare different experiments and more easily reproduce results.

Installation

pip install git+https://github.com/martno/exprec.git

Usage

Code example

from exprec import Experiment

with Experiment() as experiment:
    pass
    # YOUR CODE HERE

All output in this with statement will be logged to a file which can later be accessed in the dashboard (see below). Also, all python code and installed packages are recorded as well.

Dashboard

In your terminal:

exprec

Now visit http://localhost:5000/ in your browser to see the dashboard.

More code examples

from exprec import Experiment

with Experiment(name='experiment-1', tags=['tag1', 'tag2']) as experiment:
    experiment.set_parameter('test_parameter', 5)

    experiment.add_scalar('scalar1', 4, step=0)
    experiment.add_scalar('scalar1', 5, step=1)

    with experiment.open('filename.txt', mode='w') as fp:
        fp.write('test\n')
        # This creates a file in the experiment's folder (`.experiments/<experiment-id>/files/filename.txt`). It can be
        # accessed by other experiments using
        # `with experiment.open('filename.txt', mode='w', uuid=previous_experiment_uuid) as fp:`.
        # When opening a previous experiment's file, the previous experiment will be referred to as 
        # the current experiment's parent (shown in the summary in the dashboard). 

    raise ValueError('Invalid value')
    # The experiment will finish with status 'failed'. The exception is also logged. 

Licence

MIT Licence

About

Exprec records your execution runs so you can compare different experiments and more easily reproduce results

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 73.3%
  • JavaScript 14.9%
  • HTML 10.7%
  • CSS 1.1%