lllyasviel/ControlNet · error · ValueError

Invalid epsilon value: {}

Error message

Invalid epsilon value: {}

What it means

Raised by the EMA-tracking AdamW optimizer in ldm/util.py when the epsilon parameter is negative (eps < 0.0). eps is the numerical-stability term added to denominators in Adam-style updates; any negative value aborts optimizer construction, matching torch.optim.AdamW's validation.

Source

Thrown at ldm/util.py:99

def get_obj_from_str(string, reload=False):
    module, cls = string.rsplit(".", 1)
    if reload:
        module_imp = importlib.import_module(module)
        importlib.reload(module_imp)
    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)

View on GitHub (pinned to ed85cd1e25)

Solutions

1. Fix the sign: eps should be a small positive float like 1e-8
2. Validate config values before constructing the optimizer (assert eps > 0)
3. Sanity-check the hyperparameter dict with a small lint step in training scripts

Example fix

# before
opt = AdamW(params, eps=-1e-8)
# after
opt = AdamW(params, eps=1e-8)

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

When it happens

Trigger: Constructing AdamW(params, eps=-1e-8) or pulling eps from a config where it was mistyped or computed as negative.

Common situations: Copy-pasted hyperparameter blocks with sign typos, config generation scripts that subtract instead of add, or argparse values entered with a stray minus.

Related errors

• Invalid learning rate: {}
• Invalid beta parameter at index 0: {}
• Invalid beta parameter at index 1: {}
• Invalid weight_decay value: {}
• Invalid ema_decay value: {}

AI-assisted analysis of lllyasviel/ControlNet@ed85cd1e25 (2026-08-27). Data as JSON: /api/errors/b48fbd16dea43477. Report an issue: GitHub.