facebookresearch/detectron2 · error · ValueError

Milestones should be a list of increasing integers. Got {}

Error message

Milestones should be a list of increasing integers. Got {}

What it means

WarmupMultiStepLR (deprecated) requires milestones to be a strictly increasing list of ints; an unsorted or duplicated list fails the sorted() equality check in __init__.

Source

Thrown at detectron2/solver/lr_scheduler.py:156

# MultiStepLR with WarmupLR but the current LRScheduler design doesn't allow it.


class WarmupMultiStepLR(LRScheduler):
    def __init__(
        self,
        optimizer: torch.optim.Optimizer,
        milestones: List[int],
        gamma: float = 0.1,
        warmup_factor: float = 0.001,
        warmup_iters: int = 1000,
        warmup_method: str = "linear",
        last_epoch: int = -1,
    ):
        logger.warning(
            "WarmupMultiStepLR is deprecated! Use LRMultipilier with fvcore ParamScheduler instead!"
        )
        if not list(milestones) == sorted(milestones):
            raise ValueError(
                "Milestones should be a list of" " increasing integers. Got {}", milestones
            )
        self.milestones = milestones
        self.gamma = gamma
        self.warmup_factor = warmup_factor
        self.warmup_iters = warmup_iters
        self.warmup_method = warmup_method
        super().__init__(optimizer, last_epoch)

    def get_lr(self) -> List[float]:
        warmup_factor = _get_warmup_factor_at_iter(
            self.warmup_method, self.last_epoch, self.warmup_iters, self.warmup_factor
        )
        return [
            base_lr * warmup_factor * self.gamma ** bisect_right(self.milestones, self.last_epoch)
            for base_lr in self.base_lrs
        ]

View on GitHub (pinned to a2f4a8771a)

Solutions

1. Sort milestones ascending and remove duplicates before constructing
2. Fix cfg.SOLVER.STEPS = [10000, 30000] ordering in the config
3. Migrate to LRMultiplier with fvcore MultiStepParamScheduler

Example fix

# before
cfg.SOLVER.STEPS = (30000, 10000)
# after
cfg.SOLVER.STEPS = (10000, 30000)

steps = list(cfg.SOLVER.STEPS)
assert steps == sorted(steps) and len(set(steps)) == len(steps), 'SOLVER.STEPS must be increasing, unique'

When it happens

Trigger: Passing milestones=[30000, 10000] or [1000, 1000, 20000] to WarmupMultiStepLR, typically via cfg.SOLVER.STEPS in wrong order.

Common situations: Hand-edited YAML configs where steps were appended out of order; merging config files that reorder SOLVER.STEPS.

Understand the failure class

Background: "Invalid value" and "allowed values are" config errors: what your library rejected and how to fix it — this error's family across 41 libraries.

Related errors

• Unknown LR scheduler: {}
• Unknown warmup method: {}
• Class with @configurable must have a 'from_config' classmeth
• {name} must take 'cfg' as the first argument!
• total_batch_size and single_gpu_batch_size are mutually inco

AI-assisted analysis of facebookresearch/detectron2@a2f4a8771a (2026-08-27). Data as JSON: /api/errors/d2fbdba6a678f45f. Report an issue: GitHub.