open-mmlab/mmdetection · error · ValueError

batch_size should be a positive integer value, but got batch

Error message

batch_size should be a positive integer value, but got batch_size={batch_size}

What it means

AspectRatioBatchSampler validates that batch_size is a positive Python int (bools excluded implicitly by the isinstance check combined with <=0 guard only for numbers; actually bool passes isinstance(int)). A non-int or non-positive value triggers this ValueError at construction.

Source

Thrown at mmdet/datasets/samplers/batch_sampler.py:32

    >= 1) into a same batch.

    Args:
        sampler (Sampler): Base sampler.
        batch_size (int): Size of mini-batch.
        drop_last (bool): If ``True``, the sampler will drop the last batch if
            its size would be less than ``batch_size``.
    """

    def __init__(self,
                 sampler: Sampler,
                 batch_size: int,
                 drop_last: bool = False) -> None:
        if not isinstance(sampler, Sampler):
            raise TypeError('sampler should be an instance of ``Sampler``, '
                            f'but got {sampler}')
        if not isinstance(batch_size, int) or batch_size <= 0:
            raise ValueError('batch_size should be a positive integer value, '
                             f'but got batch_size={batch_size}')
        self.sampler = sampler
        self.batch_size = batch_size
        self.drop_last = drop_last
        # two groups for w < h and w >= h
        self._aspect_ratio_buckets = [[] for _ in range(2)]

    def __iter__(self) -> Sequence[int]:
        for idx in self.sampler:
            data_info = self.sampler.dataset.get_data_info(idx)
            width, height = data_info['width'], data_info['height']
            bucket_id = 0 if width < height else 1
            bucket = self._aspect_ratio_buckets[bucket_id]
            bucket.append(idx)
            # yield a batch of indices in the same aspect ratio group
            if len(bucket) == self.batch_size:
                yield bucket[:]
                del bucket[:]

View on GitHub (pinned to cfd5d3a985)

Solutions

1. Convert to a plain positive int before passing: batch_size=int(batch_size)
2. Ensure the computed value is >= 1 (guard divisions like len(dataset)//world_size so they never yield 0)
3. Fix the config so batch_size is an integer literal, not a string or float

Example fix

# before
batch_sampler = AspectRatioBatchSampler(sampler=s, batch_size='4')
# after
batch_sampler = AspectRatioBatchSampler(sampler=s, batch_size=4)

assert isinstance(batch_size, int) and not isinstance(batch_size, bool) and batch_size > 0, 'batch_size must be a positive int'

def is_positive_int(v) -> bool:
    return isinstance(v, int) and not isinstance(v, bool) and v > 0

When it happens

Trigger: Passing batch_size as a float (2.0), a string ('2'), 0, a negative number, or None to AspectRatioBatchSampler; commonly happens when batch_size comes from a config/env variable parsed as string or float.

Common situations: Reading batch size from CLI args or environment without int() conversion; computing batch_size dynamically (e.g. len(dataset)//num_gpus yielding 0 for tiny datasets); config templating that leaves it as a string.

Understand the failure class

Background: "Must be a positive integer", "Invalid value", "Unsupported": the invalid-argument-value error family, when a library rejects the value you pass — this error's family across 35 libraries.

Related errors

• sampler should be an instance of ``Sampler``, but got {sampl
• dataset metainfo must contain `classes`
• img_border_value must be float or tuple with 3 elements.
• metric item "{metric_item}" is not supported
• metric must be a list or a str.

AI-assisted analysis of open-mmlab/mmdetection@cfd5d3a985 (2026-08-27). Data as JSON: /api/errors/2c1f78bb47870f68. Report an issue: GitHub.