lllyasviel/ControlNet · error · ValueError

Invalid weight_decay value: {}

Error message

Invalid weight_decay value: {}

What it means

Raised by the EMA-tracking AdamW optimizer when weight_decay is negative (weight_decay < 0.0). This optimizer (unlike some SGD variants) only accepts zero or positive weight decay; a negative value aborts construction.

Source

Thrown at ldm/util.py:105

    return getattr(importlib.import_module(module, package=None), cls)


class AdamWwithEMAandWings(optim.Optimizer):
    # credit to https://gist.github.com/crowsonkb/65f7265353f403714fce3b2595e0b298
    def __init__(self, params, lr=1.e-3, betas=(0.9, 0.999), eps=1.e-8,  # TODO: check hyperparameters before using
                 weight_decay=1.e-2, amsgrad=False, ema_decay=0.9999,   # ema decay to match previous code
                 ema_power=1., param_names=()):
        """AdamW that saves EMA versions of the parameters."""
        if not 0.0 <= lr:
            raise ValueError("Invalid learning rate: {}".format(lr))
        if not 0.0 <= eps:
            raise ValueError("Invalid epsilon value: {}".format(eps))
        if not 0.0 <= betas[0] < 1.0:
            raise ValueError("Invalid beta parameter at index 0: {}".format(betas[0]))
        if not 0.0 <= betas[1] < 1.0:
            raise ValueError("Invalid beta parameter at index 1: {}".format(betas[1]))
        if not 0.0 <= weight_decay:
            raise ValueError("Invalid weight_decay value: {}".format(weight_decay))
        if not 0.0 <= ema_decay <= 1.0:
            raise ValueError("Invalid ema_decay value: {}".format(ema_decay))
        defaults = dict(lr=lr, betas=betas, eps=eps,
                        weight_decay=weight_decay, amsgrad=amsgrad, ema_decay=ema_decay,
                        ema_power=ema_power, param_names=param_names)
        super().__init__(params, defaults)

    def __setstate__(self, state):
        super().__setstate__(state)
        for group in self.param_groups:
            group.setdefault('amsgrad', False)

    @torch.no_grad()
    def step(self, closure=None):
        """Performs a single optimization step.
        Args:
            closure (callable, optional): A closure that reevaluates the model
                and returns the loss.

View on GitHub (pinned to ed85cd1e25)

Solutions

  1. Fix the sign: use a small non-negative value like 0.01 or 0.0 to disable
  2. If using a sweep, bound weight_decay to [0.0, 1e-1]
  3. Add pre-construction validation of the optimizer kwargs dict

Example fix

# before
opt = AdamW(params, weight_decay=-1e-2)
# after
opt = AdamW(params, weight_decay=1e-2)
Defensive patterns

Strategy: validation

Validate before calling

assert 0.0 <= weight_decay, f'bad weight_decay: {weight_decay}'

Prevention

When it happens

Trigger: Constructing AdamW(params, weight_decay=-1e-2), or configs where weight decay was entered with a minus sign intending a larger effective lr (a misunderstanding of decoupled weight decay).

Common situations: Sign typos in YAML/JSON training configs, hyperparameter search sampling negative values, or porting configs from optimizers that interpret weight decay differently.

Related errors

AI-assisted analysis of lllyasviel/ControlNet@ed85cd1e25 (2026-08-27). Data as JSON: /api/errors/6ad2793b60d144ee. Report an issue: GitHub.