MCPcopy Create free account
hub / github.com/Audio-AGI/AudioSep / setup

Method setup

data/datamodules.py:43–52  ·  view source on GitHub ↗

r"""called on every device.

(self, stage: Optional[str] = None)

Source from the content-addressed store, hash-verified

41 pass
42
43 def setup(self, stage: Optional[str] = None) -> NoReturn:
44 r"""called on every device."""
45
46 # make assignments here (val/train/test split)
47 # called on every process in DDP
48
49 # SegmentSampler is used for selecting segments for training.
50 # On multiple devices, each SegmentSampler samples a part of mini-batch
51 # data.
52 self.train_dataset = self._train_dataset
53
54
55 def train_dataloader(self) -> torch.utils.data.DataLoader:

Callers

nothing calls this directly

Calls

no outgoing calls

Tested by

no test coverage detected