Lightning-AI/pytorch-lightning · error · ValueError

You requested to find {num_devices} devices but this machine

Error message

You requested to find {num_devices} devices but this machine only has {len(visible_devices)} GPUs.

What it means

The Trainer profiler parameter accepts a Profiler instance or a string. Strings are lowercased and looked up in a fixed registry (simple, advanced, pytorch, xla). Any string not in PROFILERS raises MisconfigurationException listing the allowed keys.

Source

Thrown at src/lightning/fabric/accelerators/cuda.py:110

    Args:
        num_devices: The number of devices you want to request. By default, this function will return as many as there
            are usable CUDA GPU devices available.

    Warning:
        If multiple processes call this function at the same time, there can be race conditions in the case where
        both processes determine that the device is unoccupied, leading into one of them crashing later on.

    """
    if num_devices == 0:
        return []
    visible_devices = _get_all_visible_cuda_devices()
    if not visible_devices:
        raise ValueError(
            f"You requested to find {num_devices} devices but there are no visible CUDA devices on this machine."
        )
    if num_devices > len(visible_devices):
        raise ValueError(
            f"You requested to find {num_devices} devices but this machine only has {len(visible_devices)} GPUs."
        )

    available_devices = []
    unavailable_devices = []

    for gpu_idx in visible_devices:
        try:
            torch.tensor(0, device=torch.device("cuda", gpu_idx))
        except RuntimeError:
            unavailable_devices.append(gpu_idx)
            continue

        available_devices.append(gpu_idx)
        if len(available_devices) == num_devices:
            # exit early if we found the right number of GPUs
            break

View on GitHub (pinned to 9fed5c27d2)

Solutions

1. Use one of the registered names: 'simple', 'advanced', 'pytorch', 'xla' (lowercase)
2. Pass a Profiler instance for anything custom: from lightning.pytorch.profilers import PyTorchProfiler; Trainer(profiler=PyTorchProfiler())
3. Omit profiler (or profiler=None) to disable profiling

Example fix

# before
trainer = Trainer(profiler="PyTorch")

# after
trainer = Trainer(profiler="pytorch")
# or
from lightning.pytorch.profilers import PyTorchProfiler
trainer = Trainer(profiler=PyTorchProfiler())

ALLOWED_PROFILERS = {"simple", "advanced", "pytorch", "xla"}

def valid_profiler(p):
    if isinstance(p, str):
        assert p.lower() in ALLOWED_PROFILERS, f"profiler must be one of {ALLOWED_PROFILERS}"
    return p

from lightning.pytorch.profilers import Profiler

ALLOWED = {"simple", "advanced", "pytorch", "xla"}

def is_valid_profiler(p) -> bool:
    return p is None or isinstance(p, Profiler) or (isinstance(p, str) and p.lower() in ALLOWED)

When it happens

Trigger: Trainer(profiler='tensorboard'), Trainer(profiler='PyTorchProfiler'), Trainer(profiler='') or any misspelled/unregistered profiler name.

Common situations: Typos and case/spelling mistakes; assuming a profiler exists that this Lightning version doesn't register; passing the class name string instead of the class; older/newer versions where the available profiler set differs.

Understand the failure class

Background: Invalid enum value errors: "Unknown type", "Invalid scope", "must be one of" — when a string is not on the library's allowed list — this error's family across 23 libraries.

Related errors

• Found sort_by_key: {self._sort_by_key}. Should be within {se
• Found invalid table_kwargs key: {key}. This is already a pos
• You selected an invalid strategy name: `strategy={strategy!r
• You selected an invalid accelerator name: `accelerator={acce
• f"`Trainer(barebones=True, profiler={profiler!r})` was passe

AI-assisted analysis of Lightning-AI/pytorch-lightning@9fed5c27d2 (2026-08-28). Data as JSON: /api/errors/d5f307672b7fdd01. Report an issue: GitHub.