Share Status and Outcomes of Experiments Within the Team
Intent
Facilitate knowledge transfer, peer review and model assessment.
Motivation
Team members have different ways of managing and logging experiment related data. Adopting a common way to log experiment data and share it within the team enables members to collectively monitor and assess training outcomes.
Applicability
Experiment tracking and sharing should be used for any training experiment.
Description
Although different team members have their own style of managing experiments and tracing their outcomes, it is recommended to adopt a common way of logging data; that is understood and accessible to all team members.
Sharing the outcomes within the team has several benefits for peer review, knowledge transfer and model assessment.
Several collaborative tools enable central logging of experimental results.
Whenever possible, it is recommended to use one of the tools available internally or externally (e.g. Sacred or W&B).
Adoption
Related
Read more
- 10 Best Practices for Deep Learning
- Principled Machine Learning: Practices and Tools for Efficient Collaboration