invoke-ai/InvokeAI · error · ValueError

Invalid CFG scale type: {type(self.cfg_scale)}

Error message

Invalid CFG scale type: {type(self.cfg_scale)}

What it means

`_prepare_cfg_scale` only accepts float or list[float] for cfg_scale. Any other type (int, str, dict, etc.) reaches the final fallback raise and produces this ValueError naming the offending Python type.

Source

Thrown at invokeai/app/invocations/krea2_denoise.py:215

            KREA2_LATENT_CHANNELS,
            int(height) // LATENT_SCALE_FACTOR,
            int(width) // LATENT_SCALE_FACTOR,
            device=rand_device,
            dtype=torch.float32,
            generator=torch.Generator(device=rand_device).manual_seed(seed),
        ).to(device=device, dtype=dtype)

    def _prepare_cfg_scale(self, num_timesteps: int) -> list[float]:
        if isinstance(self.cfg_scale, float):
            return [self.cfg_scale] * num_timesteps
        if isinstance(self.cfg_scale, list):
            if len(self.cfg_scale) != num_timesteps:
                raise ValueError(
                    f"cfg_scale list has {len(self.cfg_scale)} values but the model is configured for "
                    f"{num_timesteps} steps. Provide one CFG value per configured step (or a single float)."
                )
            return self.cfg_scale
        raise ValueError(f"Invalid CFG scale type: {type(self.cfg_scale)}")

    @staticmethod
    def _should_apply_cfg_for_step(cfg_scale: float, *, has_negative_conditioning: bool) -> bool:
        return has_negative_conditioning and cfg_scale > 1.0

    @staticmethod
    def _validate_effective_schedule(*, start_idx: int, end_idx: int) -> None:
        if end_idx <= start_idx:
            raise ValueError(
                "The requested denoising range does not contain any effective denoising steps at the configured "
                "step count. Increase denoising_end, decrease denoising_start, or increase steps."
            )

    def _validate_inputs(self) -> None:
        if self.denoising_start >= self.denoising_end:
            raise ValueError("denoising_start must be less than denoising_end.")
        if self.denoise_mask is not None and self.latents is None:
            raise ValueError("Initial latents are required when a denoise mask is provided.")

View on GitHub (pinned to 0b6a024f2f)

Solutions

1. Convert cfg_scale to float before invoking: cfg_scale=float(value).
2. If passing a schedule, ensure it is a list of floats with length == steps.
3. Check the caller/serialization layer for values that skip pydantic validation (raw dict construction).

Example fix

// before
cfg_scale="3.5"          # str from JSON
cfg_scale=4              # int
// after
cfg_scale=float("3.5")   # 3.5
cfg_scale=4.0

if not (isinstance(cfg_scale, float) or (isinstance(cfg_scale, list) and all(isinstance(v, float) for v in cfg_scale))):
    raise TypeError(f"cfg_scale must be float or list[float], got {type(cfg_scale)}")

def is_valid_cfg_type(cfg_scale) -> bool:
    if isinstance(cfg_scale, float):
        return True
    return isinstance(cfg_scale, list) and all(isinstance(v, float) for v in cfg_scale)

try:
    out = invoke_krea2_denoise(cfg_scale=cfg_scale)
except (ValueError, TypeError) as e:
    if "Invalid CFG scale type" in str(e):
        cfg_scale = float(cfg_scale)
        out = invoke_krea2_denoise(cfg_scale=cfg_scale)
    else:
        raise

When it happens

Trigger: Passing cfg_scale as an int (e.g. cfg_scale=4 rather than 4.0 from a non-coercing caller), a string from JSON deserialization, or None from an unbound input field.

Common situations: Programmatic invocation construction passing raw JSON values without type coercion; UI integrations sending strings; custom scripts passing ints because pydantic coercion is bypassed.

Related errors

• cfg_scale list has {len(self.cfg_scale)} values but the mode
• Expected ModelPatchRaw for LoRA '{lora.lora.key}', got {type
• Expected PreTrainedModel for Gemma encoder, got {type(gemma_
• Expected PreTrainedTokenizerBase for Gemma tokenizer, got {t
• Expected PidNet for PiD decoder, got {type(pid_net).__name__

AI-assisted analysis of invoke-ai/InvokeAI@0b6a024f2f (2026-08-29). Data as JSON: /api/errors/c504086257aadefc. Report an issue: GitHub.