Lightning-AI/pytorch-lightning · error · MisconfigurationException

Trainer was configured with `enable_checkpointing=False` but

Error message

Trainer was configured with `enable_checkpointing=False` but found `ModelCheckpoint` in callbacks list.

What it means

Raised during Trainer initialization when enable_checkpointing=False is set but a ModelCheckpoint callback is present in the callbacks list. These are contradictory instructions, and Lightning treats the conflict as a user error instead of silently dropping the callback or overriding the flag.

Source

Thrown at src/lightning/pytorch/trainer/connectors/callback_connector.py:94

        self._configure_timer_callback(max_time)

        # init progress bar
        self._configure_progress_bar(enable_progress_bar)

        # configure the ModelSummary callback
        self._configure_model_summary_callback(enable_model_summary)

        self.trainer.callbacks.extend(_load_external_callbacks("lightning.pytorch.callbacks_factory"))
        _validate_callbacks_list(self.trainer.callbacks)

        # push all model checkpoint callbacks to the end
        # it is important that these are the last callbacks to run
        self.trainer.callbacks = self._reorder_callbacks(self.trainer.callbacks)

    def _configure_checkpoint_callbacks(self, enable_checkpointing: bool) -> None:
        if self.trainer.checkpoint_callbacks:
            if not enable_checkpointing:
                raise MisconfigurationException(
                    "Trainer was configured with `enable_checkpointing=False`"
                    " but found `ModelCheckpoint` in callbacks list."
                )
        elif enable_checkpointing:
            if RequirementCache("litmodels >=0.1.7") and self.trainer._model_registry:
                trainer_source = inspect.getmodule(self.trainer)
                if trainer_source is None or not isinstance(trainer_source.__package__, str):
                    raise RuntimeError("Unable to determine the source of the trainer.")
                # this need to imported based on the actual package lightning/pytorch_lightning
                if "pytorch_lightning" in trainer_source.__package__:
                    from litmodels.integrations.checkpoints import PytorchLightningModelCheckpoint as LitModelCheckpoint
                else:
                    from litmodels.integrations.checkpoints import LightningModelCheckpoint as LitModelCheckpoint

                model_checkpoint = LitModelCheckpoint(model_registry=self.trainer._model_registry)
            else:
                # Defer the litmodels tip until loggers are set up (in _attach_model_callbacks)
                self._pending_litmodels_tip = True

View on GitHub (pinned to 9fed5c27d2)

Solutions

1. Remove the ModelCheckpoint callback from callbacks when using enable_checkpointing=False
2. Set enable_checkpointing=True (or omit it, default True) and configure ModelCheckpoint(...) with the desired dirpath/filename/monitor
3. Filter the callback list programmatically before constructing the Trainer

Example fix

# before
trainer = Trainer(enable_checkpointing=False, callbacks=[ModelCheckpoint(dirpath="ckpts")])
# after
trainer = Trainer(enable_checkpointing=False, callbacks=[])
# or keep checkpointing on:
trainer = Trainer(callbacks=[ModelCheckpoint(dirpath="ckpts", monitor="val_loss")])

from lightning.pytorch.callbacks import ModelCheckpoint

def build_callbacks(enable_checkpointing: bool, extra):
    cbs = list(extra)
    if not enable_checkpointing:
        cbs = [c for c in cbs if not isinstance(c, ModelCheckpoint)]
    return cbs

trainer = Trainer(enable_checkpointing=False, callbacks=build_callbacks(False, my_callbacks))

def checkpointing_is_consistent(enable_checkpointing: bool, callbacks) -> bool:
    from lightning.pytorch.callbacks import ModelCheckpoint
    has_ckpt_cb = any(isinstance(c, ModelCheckpoint) for c in callbacks)
    return not (has_ckpt_cb and not enable_checkpointing)

When it happens

Trigger: Calling Trainer(enable_checkpointing=False, callbacks=[ModelCheckpoint(...)]) or callbacks=[..., ModelCheckpoint()] while the flag disables default checkpointing; commonly happens when a shared callback list from another trainer is reused.

Common situations: Disabling checkpointing for a quick debug run while forgetting a ModelCheckpoint added earlier; copy-pasting trainer configs that include both the flag and the callback; libraries that inject ModelCheckpoint into the callbacks list.

Related errors

• Trainer was configured with `enable_progress_bar=False` but
• Found more than one stateful callback of type `{type(callbac
• You added multiple progress bar callbacks to the Trainer, bu
• Filter should be a dictionary, given {filter!r}
• The filter keys {filter.keys() - state} are not present in t

AI-assisted analysis of Lightning-AI/pytorch-lightning@9fed5c27d2 (2026-08-28). Data as JSON: /api/errors/4197b7c0e23e7abd. Report an issue: GitHub.