Lightning-AI/pytorch-lightning · error · MisconfigurationException

accelerator set through both strategy class and accelerator

Error message

accelerator set through both strategy class and accelerator flag, choose one

What it means

When a Strategy instance is passed to Trainer(strategy=...) that already carries an accelerator (strategy._accelerator set), you must not also set the accelerator flag explicitly. The connector refuses ambiguous accelerator ownership.

Source

Thrown at src/lightning/pytorch/trainer/connectors/accelerator_connector.py:273

                    f"Received multiple values for {', '.join(duplicated_plugin_key)} flags in `plugins`."
                    " Expected one value for each type at most."
                )

            if plugins_flags_types.get(Precision.__name__) and precision_flag is not None:
                raise ValueError(
                    f"Received both `precision={precision_flag}` and `plugins={self._precision_plugin_flag}`."
                    f" Choose one."
                )

        self._precision_flag = "32-true" if precision_flag is None else precision_flag

        # handle the case when the user passes in a strategy instance which has an accelerator, precision,
        # checkpoint io or cluster env set up
        # TODO: improve the error messages below
        if self._strategy_flag and isinstance(self._strategy_flag, Strategy):
            if self._strategy_flag._accelerator:
                if self._accelerator_flag != "auto":
                    raise MisconfigurationException(
                        "accelerator set through both strategy class and accelerator flag, choose one"
                    )
                self._accelerator_flag = self._strategy_flag._accelerator
            if self._strategy_flag._precision_plugin:
                # [RFC] handle precision plugin set up conflict?
                if self._precision_plugin_flag:
                    raise MisconfigurationException("precision set through both strategy class and plugins, choose one")
                self._precision_plugin_flag = self._strategy_flag._precision_plugin
            if self._strategy_flag._checkpoint_io:
                if self.checkpoint_io:
                    raise MisconfigurationException(
                        "checkpoint_io set through both strategy class and plugins, choose one"
                    )
                self.checkpoint_io = self._strategy_flag._checkpoint_io
            if getattr(self._strategy_flag, "cluster_environment", None):
                if self._cluster_environment_flag:
                    raise MisconfigurationException(
                        "cluster_environment set through both strategy class and plugins, choose one"

View on GitHub (pinned to 9fed5c27d2)

Solutions

1. Drop the accelerator= argument from Trainer and let the strategy provide it
2. Or construct the strategy without an accelerator and pass accelerator= to Trainer

Example fix

# before
strategy = DDPStrategy(accelerator=CUDAAccelerator())
trainer = Trainer(strategy=strategy, accelerator="gpu")
# after
strategy = DDPStrategy()
trainer = Trainer(strategy=strategy, accelerator="gpu")

if isinstance(strategy, Strategy) and strategy._accelerator and accelerator != "auto":
    accelerator = "auto"  # or raise a clear config error early
trainer = Trainer(strategy=strategy, accelerator=accelerator)

def strategy_accelerator_conflict(strategy, accelerator) -> bool:
    from lightning.pytorch.strategies import Strategy
    return isinstance(strategy, Strategy) and strategy._accelerator is not None and accelerator not in (None, "auto")

When it happens

Trigger: strategy = DDPStrategy(accelerator=CUDAAccelerator()); Trainer(strategy=strategy, accelerator='gpu') — any non-'auto' accelerator flag combined with a pre-configured strategy instance.

Common situations: Reusing a pre-configured strategy object built by a shared factory and adding accelerator='gpu' at Trainer construction; copy-pasted configs from examples that build strategies manually.

Related errors

• precision set through both strategy class and plugins, choos
• checkpoint_io set through both strategy class and plugins, c
• cluster_environment set through both strategy class and plug
• accelerator set through both strategy class and accelerator
• CPU parallel_devices set through {self._strategy_flag.__clas

AI-assisted analysis of Lightning-AI/pytorch-lightning@9fed5c27d2 (2026-08-28). Data as JSON: /api/errors/d9f4abfc40570888. Report an issue: GitHub.