microsoft/VibeVoice · error · ValueError

Number of inference steps is 'None', you need to run 'set_ti

Error message

Number of inference steps is 'None', you need to run 'set_timesteps' after creating the scheduler

What it means

`step()` requires a prepared sigma/timestep grid; `self.num_inference_steps` is None until `set_timesteps()` has run, so stepping first raises this ValueError. The scheduler deliberately stores no default schedule — every sampling loop must call set_timesteps (with a step count or custom timesteps) after construction and before the first step().

Source

Thrown at vibevoice/schedule/dpm_solver.py:970

                The current discrete timestep in the diffusion chain.
            sample (`torch.Tensor`):
                A current instance of a sample created by the diffusion process.
            generator (`torch.Generator`, *optional*):
                A random number generator.
            variance_noise (`torch.Tensor`):
                Alternative to generating noise with `generator` by directly providing the noise for the variance
                itself. Useful for methods such as [`LEdits++`].
            return_dict (`bool`):
                Whether or not to return a [`~schedulers.scheduling_utils.SchedulerOutput`] or `tuple`.

        Returns:
            [`~schedulers.scheduling_utils.SchedulerOutput`] or `tuple`:
                If return_dict is `True`, [`~schedulers.scheduling_utils.SchedulerOutput`] is returned; otherwise, a
                tuple is returned where the first element is the sample tensor.

        """
        if self.num_inference_steps is None:
            raise ValueError(
                "Number of inference steps is 'None', you need to run 'set_timesteps' after creating the scheduler"
            )

        if self.step_index is None:
            self._init_step_index(timestep)

        # Improve numerical stability for small number of steps
        lower_order_final = (self.step_index == len(self.timesteps) - 1) and (
            self.config.euler_at_final
            or (self.config.lower_order_final and len(self.timesteps) < 15)
            or self.config.final_sigmas_type == "zero"
        )
        lower_order_second = (
            (self.step_index == len(self.timesteps) - 2) and self.config.lower_order_final and len(self.timesteps) < 15
        )

        model_output = self.convert_model_output(model_output, sample=sample)
        for i in range(self.config.solver_order - 1):

View on GitHub (pinned to 94da20d98b)

Solutions

1. Call `scheduler.set_timesteps(num_inference_steps=N)` once after creating the scheduler and before the sampling loop.
2. If using custom timesteps: `scheduler.set_timesteps(timesteps=[...])`.
3. Add a guard in your loop: `if scheduler.num_inference_steps is None: scheduler.set_timesteps(30)`.

Example fix

# before
for t in scheduler.timesteps:
    scheduler.step(model_output, t, sample)  # num_inference_steps is None

# after
scheduler.set_timesteps(num_inference_steps=30)
for t in scheduler.timesteps:
    scheduler.step(model_output, t, sample)

if scheduler.num_inference_steps is None:
    scheduler.set_timesteps(num_inference_steps=30)
for t in scheduler.timesteps:
    scheduler.step(model_output, t, sample)

def scheduler_is_ready(scheduler) -> bool:
    return scheduler.num_inference_steps is not None and scheduler.timesteps is not None

When it happens

Trigger: `scheduler.step(model_output, t, sample)` before any `scheduler.set_timesteps(30)` call; also after re-creating or re-loading a scheduler mid-loop, or when an exception earlier in the pipeline skipped the set_timesteps line.

Common situations: Reordering pipeline code so the denoise loop runs first; wrapping schedulers in objects that lazy-init; copying example code that omitted the set_timesteps line; resuming from a checkpoint without re-running setup.

Related errors

• Must pass exactly one of `num_inference_steps` or `timesteps
• Can only pass one of `num_inference_steps` or `custom_timest
• missing `sample` as a required keyword argument
• missing `sample` as a required keyword argument
• missing`sample` as a required keyword argument

AI-assisted analysis of microsoft/VibeVoice@94da20d98b (2026-08-15). Data as JSON: /api/errors/0d541b3936a91565. Report an issue: GitHub.