Lightning-AI/pytorch-lightning · error · ValueError

Received multiple values for {', '.join(duplicated_plugin_ke

Error message

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

What it means

Raised by Lightning Fabric's connector when the `plugins` argument receives more than one plugin of the same category (e.g. two Precision plugins or two CheckpointIO plugins). Fabric allows at most one plugin per type: Precision, CheckpointIO, and ClusterEnvironment. The duplicate check counts plugin class names and fails when any count exceeds 1.

Source

Thrown at src/lightning/fabric/connector.py:247

            for plugin in plugins:
                if isinstance(plugin, Precision):
                    self._precision_instance = plugin
                    plugins_flags_types[Precision.__name__] += 1
                elif isinstance(plugin, CheckpointIO):
                    self.checkpoint_io = plugin
                    plugins_flags_types[CheckpointIO.__name__] += 1
                elif isinstance(plugin, ClusterEnvironment):
                    self._cluster_environment_flag = plugin
                    plugins_flags_types[ClusterEnvironment.__name__] += 1
                else:
                    raise TypeError(
                        f"Found invalid type for plugin {plugin}. Expected one of: Precision, "
                        "CheckpointIO, ClusterEnvironment."
                    )

            duplicated_plugin_key = [k for k, v in plugins_flags_types.items() if v > 1]
            if duplicated_plugin_key:
                raise ValueError(
                    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_input is not None:
                raise ValueError(
                    f"Received both `precision={precision_input}` and `plugins={self._precision_instance}`. Choose one."
                )

        self._precision_input = "32-true" if precision_input is None else precision_input

        # 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 isinstance(self._strategy_flag, Strategy):
            if self._strategy_flag._accelerator:
                if self._accelerator_flag != "auto":
                    raise ValueError("accelerator set through both strategy class and accelerator flag, choose one")

View on GitHub (pinned to 9fed5c27d2)

Solutions

1. Inspect the `plugins` list and keep only one plugin per category (Precision, CheckpointIO, ClusterEnvironment)
2. If you need mixed precision, keep only the Precision plugin and remove redundant ones
3. Pass unrelated plugin types together (one Precision + one CheckpointIO is fine) — only same-type duplicates are rejected

Example fix

# before
fabric = Fabric(plugins=[MixedPrecision("16-mixed", device="cuda"), DeepSpeedPrecision(...)])

# after
fabric = Fabric(plugins=[MixedPrecision("16-mixed", device="cuda")])

from lightning.fabric.plugins import Precision, CheckpointIO, ClusterEnvironment

def plugin_counts_ok(plugins):
    counts = {}
    for p in plugins:
        for t in (Precision, CheckpointIO, ClusterEnvironment):
            if isinstance(p, t):
                counts[t.__name__] = counts.get(t.__name__, 0) + 1
    return all(v <= 1 for v in counts.values())

assert plugin_counts_ok(plugins), "duplicate plugin type in plugins list"

When it happens

Trigger: Calling Fabric(plugins=[MixedPrecision(...), DeepSpeedPrecision(...)]) or Fabric(plugins=[CheckpointIO(), TorchCheckpointIO()]) — i.e. passing a list to `plugins` containing two instances whose classes map to the same plugin type key.

Common situations: Copying plugin lists from different examples/tutorials into one config; upgrading configs where a precision plugin was already included and another was added; mixing DeepSpeed + MixedPrecision plugins; enabling both a custom CheckpointIO and a built-in one.

Related errors

• Received both `precision={precision_input}` and `plugins={se
• 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
• Precision {repr(precision)} is invalid. Allowed precision va

AI-assisted analysis of Lightning-AI/pytorch-lightning@9fed5c27d2 (2026-08-28). Data as JSON: /api/errors/e2ed5c898c57175a. Report an issue: GitHub.