microsoft/VibeVoice · error · ValueError

{self.config.timestep_spacing} is not supported. Please make

Error message

{self.config.timestep_spacing} is not supported. Please make sure to choose one of 'linspace', 'leading' or 'trailing'.

What it means

When `set_timesteps` generates its grid from `num_inference_steps`, the spacing rule comes from `config.timestep_spacing`; this vendored scheduler implements only `linspace`, `leading`, and `trailing`. Any other value raises ValueError at set_timesteps time (not construction), because the constructor never validates it. The default in diffusers-style configs is usually `leading`.

Source

Thrown at vibevoice/schedule/dpm_solver.py:380

                    .copy()
                    .astype(np.int64)
                )
            elif self.config.timestep_spacing == "leading":
                step_ratio = last_timestep // (num_inference_steps + 1)
                # creates integer timesteps by multiplying by ratio
                # casting to int to avoid issues when num_inference_step is power of 3
                timesteps = (
                    (np.arange(0, num_inference_steps + 1) * step_ratio).round()[::-1][:-1].copy().astype(np.int64)
                )
                timesteps += self.config.steps_offset
            elif self.config.timestep_spacing == "trailing":
                step_ratio = self.config.num_train_timesteps / num_inference_steps
                # creates integer timesteps by multiplying by ratio
                # casting to int to avoid issues when num_inference_step is power of 3
                timesteps = np.arange(last_timestep, 0, -step_ratio).round().copy().astype(np.int64)
                timesteps -= 1
            else:
                raise ValueError(
                    f"{self.config.timestep_spacing} is not supported. Please make sure to choose one of 'linspace', 'leading' or 'trailing'."
                )

        sigmas = np.array(((1 - self.alphas_cumprod) / self.alphas_cumprod) ** 0.5)
        log_sigmas = np.log(sigmas)

        if self.config.use_karras_sigmas:
            sigmas = np.flip(sigmas).copy()
            sigmas = self._convert_to_karras(in_sigmas=sigmas, num_inference_steps=num_inference_steps)
            timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas]).round()
        elif self.config.use_lu_lambdas:
            lambdas = np.flip(log_sigmas.copy())
            lambdas = self._convert_to_lu(in_lambdas=lambdas, num_inference_steps=num_inference_steps)
            sigmas = np.exp(lambdas)
            timesteps = np.array([self._sigma_to_t(sigma, log_sigmas) for sigma in sigmas]).round()
        else:
            sigmas = np.interp(timesteps, np.arange(0, len(sigmas)), sigmas)

View on GitHub (pinned to 94da20d98b)

Solutions

Set timestep_spacing to "linspace", "leading", or "trailing" in the scheduler constructor/config.
For diffusion models trained with zero-terminal-SNR, prefer "trailing".
Validate the value right after loading config, since the constructor will not catch it.

Example fix

# before
sched = DPMSolverMultistepScheduler(..., timestep_spacing="euler")
sched.set_timesteps(30)  # ValueError here

# after
sched = DPMSolverMultistepScheduler(..., timestep_spacing="trailing")
sched.set_timesteps(30)

Defensive patterns

Strategy: validation

Validate before calling

SPACINGS = {"linspace", "leading", "trailing"}
if scheduler.config.timestep_spacing not in SPACINGS:
    raise ValueError(
        f"timestep_spacing {scheduler.config.timestep_spacing!r} invalid; "
        f"choose from {sorted(SPACINGS)}"
    )
scheduler.set_timesteps(30)

Type guard

def is_supported_timestep_spacing(v) -> bool:
    return isinstance(v, str) and v in {"linspace", "leading", "trailing"}

Prevention

Validate timestep_spacing immediately after scheduler construction — the constructor does not check it.
Configs from other diffusion repos may carry spacing values this copy lacks.
For zero-terminal-SNR-trained models, standardize on 'trailing'.

When it happens

Trigger: `scheduler.set_timesteps(30)` on a scheduler constructed with `timestep_spacing="linspace_trailing"`, `"trailing" ` with whitespace, or a value from a newer diffusers config not ported here.

Common situations: Configs authored for SDXL-era diffusers (where trailing is recommended) copied into this repo; typo or case error in YAML; model-card JSON carrying an unsupported spacing value.

Related errors

AI-assisted analysis of microsoft/VibeVoice@94da20d98b (2026-08-15). Data as JSON: /api/errors/d4112326b15b72e1. Report an issue: GitHub.