Lightning-AI/pytorch-lightning · error · ValueError

{seed} is not in bounds, numpy accepts from {min_seed_value}

Error message

{seed} is not in bounds, numpy accepts from {min_seed_value} to {max_seed_value}

What it means

`seed_everything` requires the seed to be within [4294967292? no —] `min_seed_value`..`max_seed_value`, the range numpy accepts (0 to 2**32 - 1). After parsing the seed (from the argument or PL_GLOBAL_SEED) it validates the bounds and raises ValueError otherwise.

Source

Thrown at src/lightning/fabric/utilities/seed.py:54

        verbose: Whether to print a message on each rank with the seed being set.

    """
    if seed is None:
        env_seed = os.environ.get("PL_GLOBAL_SEED")
        if env_seed is None:
            seed = 0
            if verbose:
                rank_zero_warn(f"No seed found, seed set to {seed}")
        else:
            try:
                seed = int(env_seed)
            except ValueError:
                raise ValueError(f"Invalid seed specified via PL_GLOBAL_SEED: {repr(env_seed)}")
    elif not isinstance(seed, int):
        seed = int(seed)

    if not (min_seed_value <= seed <= max_seed_value):
        raise ValueError(f"{seed} is not in bounds, numpy accepts from {min_seed_value} to {max_seed_value}")

    if verbose:
        log.info(rank_prefixed_message(f"Seed set to {seed}", _get_rank()))

    os.environ["PL_GLOBAL_SEED"] = str(seed)
    random.seed(seed)
    if _NUMPY_AVAILABLE:
        import numpy as np

        np.random.seed(seed)
    torch.manual_seed(seed)

    os.environ["PL_SEED_WORKERS"] = f"{int(workers)}"

    return seed


def reset_seed() -> None:

View on GitHub (pinned to 9fed5c27d2)

Solutions

1. Use a seed in [0, 4294967295], e.g. `seed_everything(42)`
2. If generating seeds from hashes/time, clamp: `seed = seed % (2**32)`
3. Avoid -1/None sentinels; guard configs that feed seed values

Example fix

# before
seed_everything(int(time.time()))  # or -1, or huge hash -> out of bounds

# after
seed = int(time.time()) % (2**32)
seed_everything(seed)

MIN, MAX = 0, 2**32 - 1
seed = int(seed) % (2**32)
assert MIN <= seed <= MAX

When it happens

Trigger: Calling `seed_everything(seed)` with a negative integer, a value > 4294967295, or something like `seed_everything(-1)`; a float/string argument that converts to an out-of-range int; PL_GLOBAL_SEED set to a huge number.

Common situations: Using `-1` or `None`-sentinel values as seeds, using a 64-bit hash or timestamp-derived value that overflows numpy's range, generated seeds from config sweeps exceeding 2**32-1.

Related errors

• Invalid seed specified via PL_GLOBAL_SEED: {repr(env_seed)}
• Expected samples ({samples}) to be greater or equal than bat
• Unknown configuration for model optimizers. Output from `mod
• The lr scheduler dict must have the key "scheduler" with its
• The "interval" key in lr scheduler dict must be "step" or "e

AI-assisted analysis of Lightning-AI/pytorch-lightning@9fed5c27d2 (2026-08-28). Data as JSON: /api/errors/7900b196d6c945f5. Report an issue: GitHub.