huggingface/pytorch-image-models · critical · ValueError

ScheduledBatchSampler requires a non-empty sampler.

Error message

ScheduledBatchSampler requires a non-empty sampler.

What it means

ScheduledBatchSampler rejects samplers whose __len__() is <= 0 because there would be nothing to sample; raised right after the __len__ presence check during construction.

Source

Thrown at timm/data/scheduled_sampler.py:63

    def __init__(
            self,
            sampler: Sampler,
            batch_sizes: Sequence[int],
            choice_weights: Optional[Sequence[float]] = None,
            seed: int = 0,
            drop_last: bool = True,
            shuffle_schedule: bool = True,
            num_batches: Optional[int] = None,
            choice_schedule: str = 'constant',
            schedule_epochs: Optional[int] = None,
            schedule_spread: float = 0.65,
            schedule_random_mix: float = 0.1,
    ) -> None:
        if not hasattr(sampler, '__len__'):
            raise TypeError('ScheduledBatchSampler requires a sampler with a length.')
        if len(sampler) <= 0:
            raise ValueError('ScheduledBatchSampler requires a non-empty sampler.')
        if not batch_sizes:
            raise ValueError('batch_sizes must contain at least one value.')
        if any(int(batch_size) != batch_size or batch_size <= 0 for batch_size in batch_sizes):
            raise ValueError('All scheduled batch sizes must be positive integers.')
        if num_batches is not None and (int(num_batches) != num_batches or num_batches <= 0):
            raise ValueError('num_batches must be a positive integer when specified.')
        if choice_schedule not in ('constant', 'progressive'):
            raise ValueError("choice_schedule must be 'constant' or 'progressive'.")
        if choice_schedule == 'progressive':
            if len(batch_sizes) < 2:
                raise ValueError('A progressive schedule requires at least two choices.')
            if schedule_epochs is None or int(schedule_epochs) != schedule_epochs or schedule_epochs <= 0:
                raise ValueError('schedule_epochs must be a positive integer for a progressive schedule.')
            if schedule_spread < 0:
                raise ValueError('schedule_spread must be non-negative.')
            if not 0 <= schedule_random_mix <= 1:
                raise ValueError('schedule_random_mix must be between 0 and 1.')

View on GitHub (pinned to 9a5261e31b)

Solutions

1. Check len(sampler) / len(dataset) before constructing and fix the empty data source (correct path, split, or filter).
2. For distributed runs, verify world_size divides or shards the data so each rank gets samples.
3. Log dataset size early in your training script to catch empty loads immediately.

Example fix

assert len(sampler) > 0, f'empty sampler ({len(sampler)} items) — check dataset path/filters'
sched = ScheduledBatchSampler(sampler, batch_sizes=[128, 256])

assert len(sampler) > 0, 'sampler/dataset is empty — check data loading'

When it happens

Trigger: Passing a sampler over an empty dataset (len(dataset)==0), a filtered/subset dataset with zero matching items, or a sampler initialized with n=0.

Common situations: Empty train split after a bad filter or path; Subset/DatasetFilter removing all rows; misconfigured distributed setup where the local rank got an empty shard.

Related errors

• ScheduledBatchSampler requires a sampler with a length.
• batch_sizes must contain at least one value.
• All scheduled batch sizes must be positive integers.
• num_batches must be a positive integer when specified.
• choice_schedule must be 'constant' or 'progressive'.

AI-assisted analysis of huggingface/pytorch-image-models@9a5261e31b (2026-08-27). Data as JSON: /api/errors/ab5ab8a101ef13f2. Report an issue: GitHub.