Manage ML experiments with ease.
CoordML is a light-weight Machine Learning experiment tool that enables users to easily define computational tasks and distribute tasks across multiple instances. It can also aggregate and display experiment results through a unifed web interface.
CoordML is currently working in progress. It consists of the following major components:
- Sircle DSL, a functional and interpreted language, used to define computational tasks in a expressive and flexible way. (Finished)
- CoordML Central, the central server to create experiments and distribute tasks. (Finished)
- Runner, to run tasks and extract results. (Finished)
- Web Portal, to display runner status and experiment results and aggregate experiment results in a flexible and clear way. (Finish)
Sircle is a powerful DSL for defining computational tasks. It is in functional styles, supporting first-class functions, ML-style application and currying.
A simple example of Sircle:
map (compose println $ x => "Hello, " + x) ["Sircle", "CoordML"]
This will print out
Hello, Sircle
Hello, CoordML
For a detailed guide to Sircle, you may refer to the Documentation.