Lightning-AI/pytorch-lightning · error · AttributeError

Your LightningModule code tried to access `self.trainer.{ite

Error message

Your LightningModule code tried to access `self.trainer.{item}` but this attribute is not available when using Fabric with a LightningModule.

What it means

When a LightningModule is used standalone under Fabric, `self.trainer` does not exist; attribute lookups are forwarded to the Fabric object. If Fabric also lacks the requested attribute, Lightning raises this AttributeError clarifying that `self.trainer.<item>` is not available in Fabric mode.

Source

Thrown at src/lightning/pytorch/core/module.py:1846

    LightningModule._jit_is_scripting = True
    try:
        yield
    finally:
        LightningModule._jit_is_scripting = False


class _TrainerFabricShim:
    """Intercepts attribute access on LightningModule's trainer reference and redirects it to the Fabric object."""

    def __init__(self, fabric: lf.Fabric) -> None:
        super().__init__()
        self._fabric = fabric

    def __getattr__(self, item: Any) -> Any:
        try:
            return getattr(self._fabric, item)
        except AttributeError:
            raise AttributeError(
                f"Your LightningModule code tried to access `self.trainer.{item}` but this attribute is not available"
                f" when using Fabric with a LightningModule."
            )

View on GitHub (pinned to 9fed5c27d2)

Solutions

1. Branch on availability: use `getattr`/`isinstance(self._fabric, ...)` or check `self._fabric is not None` before touching trainer attributes
2. Replace trainer lookups with Fabric equivalents (fabric.world_size, fabric.global_rank) or your own state
3. Guard shared code with `if hasattr(self, "trainer") and self.trainer is not None` style checks (careful: __getattr__ is only called when normal lookup fails)

Example fix

# before
ws = self.trainer.world_size  # under Fabric -> AttributeError
# after
fabric = self._fabric
ws = fabric.world_size if fabric is not None else self.trainer.world_size

fabric = getattr(model, "_fabric", None)
val = getattr(fabric, item, None) if fabric is not None else getattr(model.trainer, item, None)

def has_fabric(model) -> bool:
    return getattr(model, "_fabric", None) is not None

try:
    v = model.trainer.world_size
except AttributeError:
    v = model._fabric.world_size

When it happens

Trigger: Sharing a LightningModule between Trainer and Fabric workflows where the code accesses self.trainer.strategy, self.trainer.world_size, etc., and running it under Fabric.

Common situations: Migrating a Trainer-based module to Fabric while keeping trainer-specific logic, or accessing training-state attributes that only exist in the Trainer.

Related errors

• Received multiple values for {', '.join(duplicated_plugin_ke
• Received both `precision={precision_input}` and `plugins={se
• accelerator set through both strategy class and accelerator
• precision set through both strategy class and plugins, choos
• checkpoint_io set through both strategy class and plugins, c

AI-assisted analysis of Lightning-AI/pytorch-lightning@9fed5c27d2 (2026-08-28). Data as JSON: /api/errors/3ee6861ee67dd202. Report an issue: GitHub.