MCPcopy Create free account
hub / github.com/HA-Transformer/MAT / Trainer

Class Trainer

fairseq/trainer.py:24–518  ·  view source on GitHub ↗

Main class for data parallel training. This class supports synchronous distributed data parallel training, where multiple workers each have a full model replica and gradients are accumulated across workers before each update. We use :class:`~torch.nn.parallel.DistributedDataParallel

Source from the content-addressed store, hash-verified

source not stored for this graph (policy: none)

Callers 2

mainFunction · 0.90
mainFunction · 0.90

Calls

no outgoing calls

Tested by

no test coverage detected