Method
__init__
(
self,
dataset: Dataset,
weights: Sequence[float],
num_samples_per_rank: int | None = None,
generator: torch.Generator | None = None,
even_divisible: bool = True,
num_replicas: int | None = None,
rank: int | None = None,
**kwargs,
)
Source from the content-addressed store, hash-verified
| 86 | """ |
| 87 | |
| 88 | def __init__( |
| 89 | self, |
| 90 | dataset: Dataset, |
| 91 | weights: Sequence[float], |
| 92 | num_samples_per_rank: int | None = None, |
| 93 | generator: torch.Generator | None = None, |
| 94 | even_divisible: bool = True, |
| 95 | num_replicas: int | None = None, |
| 96 | rank: int | None = None, |
| 97 | **kwargs, |
| 98 | ): |
| 99 | kwargs.setdefault("shuffle", True) |
| 100 | super().__init__(dataset=dataset, even_divisible=even_divisible, num_replicas=num_replicas, rank=rank, **kwargs) |
| 101 | self.weights = weights |
| 102 | self.num_samples_per_rank = num_samples_per_rank if num_samples_per_rank is not None else self.num_samples |
| 103 | self.generator = generator |
| 104 | |
| 105 | def __iter__(self): |
| 106 | indices = list(super().__iter__()) |
Callers
nothing calls this directly
Tested by
no test coverage detected