huggingface/pytorch-image-models · error · ValueError

A progressive schedule requires at least two choices.

Error message

A progressive schedule requires at least two choices.

What it means

A progressive schedule interpolates between batch-size choices across epochs, so it needs at least two entries in batch_sizes to move between; with one (or zero) entries there is nothing to progress through and construction fails.

Source

Thrown at timm/data/scheduled_sampler.py:74

            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.')

        self.sampler = sampler
        self.batch_sizes = tuple(int(batch_size) for batch_size in batch_sizes)
        self.choice_weights = self._normalize_choice_weights(choice_weights)
        self._active_choices = tuple(
            choice_index
            for choice_index, choice_weight in enumerate(self.choice_weights)
            if choice_weight > 0
        )
        self.seed = seed
        self.drop_last = drop_last
        self.shuffle_schedule = shuffle_schedule

View on GitHub (pinned to 9a5261e31b)

Solutions

1. Provide >=2 sizes, ordered small->large, e.g. batch_sizes=[128, 256, 512].
2. If you only want one batch size, use choice_schedule='constant' with that single size.

Example fix

# before
ScheduledBatchSampler(s, batch_sizes=[256], choice_schedule='progressive')

# after
ScheduledBatchSampler(s, batch_sizes=[64, 256], choice_schedule='progressive', schedule_epochs=20)

if choice_schedule == 'progressive':
    assert len(batch_sizes) >= 2, 'progressive needs >= 2 batch sizes'

When it happens

Trigger: choice_schedule='progressive' together with batch_sizes=[256] (single element) or [].

Common situations: Configs generated from a single base batch size where the progressive path was enabled by default; forgetting to add the smaller starting batch size.

Related errors

• All scheduled batch sizes must be positive integers.
• num_batches must be a positive integer when specified.
• schedule_epochs must be a positive integer for a progressive
• schedule_spread must be non-negative.
• schedule_random_mix must be between 0 and 1.

AI-assisted analysis of huggingface/pytorch-image-models@9a5261e31b (2026-08-27). Data as JSON: /api/errors/a1c14ca18e320914. Report an issue: GitHub.