Lightning-AI/pytorch-lightning · error · ValueError

Trainer is already configured with a `BatchSizeFinder` callb

Error message

Trainer is already configured with a `BatchSizeFinder` callback.Please remove it if you want to use the Tuner.

What it means

_check_scale_batch_size_configuration rejects tuner.scale_batch_size when a BatchSizeFinder callback is already attached to the Trainer, to avoid conflicting duplicate batch-size tuning logic.

Source

Thrown at src/lightning/pytorch/tuner/tuning.py:250

    configured_callbacks = [cb for cb in trainer.callbacks if isinstance(cb, LearningRateFinder)]
    if configured_callbacks:
        raise ValueError(
            "Trainer is already configured with a `LearningRateFinder` callback."
            "Please remove it if you want to use the Tuner."
        )


def _check_scale_batch_size_configuration(trainer: "pl.Trainer") -> None:
    if trainer._accelerator_connector.is_distributed:
        raise ValueError("Tuning the batch size is currently not supported with distributed strategies.")

    # local import to avoid circular import
    from lightning.pytorch.callbacks.batch_size_finder import BatchSizeFinder

    configured_callbacks = [cb for cb in trainer.callbacks if isinstance(cb, BatchSizeFinder)]
    if configured_callbacks:
        raise ValueError(
            "Trainer is already configured with a `BatchSizeFinder` callback."
            "Please remove it if you want to use the Tuner."
        )

View on GitHub (pinned to 9fed5c27d2)

Solutions

1. Remove the BatchSizeFinder callback and use only tuner.scale_batch_size
2. Or keep the callback and drop the tuner call

Example fix

# before
trainer = Trainer(callbacks=[BatchSizeFinder()])
trainer.tuner.scale_batch_size(model)
# after
trainer = Trainer()
bs = trainer.tuner.scale_batch_size(model)

from lightning.pytorch.callbacks.batch_size_finder import BatchSizeFinder
assert not any(isinstance(cb, BatchSizeFinder) for cb in trainer.callbacks)

def has_batch_size_finder(trainer) -> bool:
    from lightning.pytorch.callbacks.batch_size_finder import BatchSizeFinder
    return any(isinstance(cb, BatchSizeFinder) for cb in trainer.callbacks)

When it happens

Trigger: trainer = Trainer(callbacks=[BatchSizeFinder(...)]) followed by trainer.tuner.scale_batch_size(model).

Common situations: Mixing the callback-based auto batch-size tuning with the Tuner API in the same script.

Related errors

• Trainer is already configured with a `LearningRateFinder` ca
• Tuning the batch size is currently not supported with distri
• Received multiple values for {', '.join(duplicated_plugin_ke
• Trainer was configured with `enable_checkpointing=False` but
• You added multiple progress bar callbacks to the Trainer, bu

AI-assisted analysis of Lightning-AI/pytorch-lightning@9fed5c27d2 (2026-08-28). Data as JSON: /api/errors/7b7d62b2a2c1ed18. Report an issue: GitHub.