pg.tuning

Distributed tuning with pluggable backends.

pg.iter provides an interface for sampling examples from a search space within a process. To support distributed tuning, PyGlove introduces pg.sample, which is almost identical but with more features:

• Allow multiple worker processes (aka. workers) to collaborate on a search with failover handling.

• Each worker can process different trials, or can cowork on the same trials via work groups.

• Provide APIs for communicating between the co-workers.

• Provide API for retrieving the search results.

• Provide a pluggable backend system for supporting user infrastructures.

Classes

• Backend
• EarlyStoppingPolicy
• Feedback
• Measurement
• RaceConditionError
• Result
• Trial

Functions

• add_backend
• available_backends
• default_backend
• poll_result
• sample
• set_default_backend