This branch introduces several new features that facilitate multi-model and multi-gpu training/validation under a single, shared dataprovider. This branch also adds support for custom training/validation routines and increased control over pre-training checks (e.g. git repo status). Note: the outward facing API/interface actually remains unchanged.
This branch introduces several new features that facilitate multi-model and multi-gpu training/validation under a single, shared dataprovider. This branch also adds support for custom training/validation routines and increased control over pre-training checks (e.g. git repo status). Note: the outward facing API/interface actually remains unchanged.