huggingface/transformers · error · ValueError

When setting ``padding={PaddingStrategy.MAX_LENGTH}``, make

Error message

When setting ``padding={PaddingStrategy.MAX_LENGTH}``, make sure that max_length is defined

What it means

PaddingStrategy.MAX_LENGTH pads everything to a fixed max_length, so that length must be known. If padding resolves to 'max_length' and max_length is None, the strategy is meaningless and the extractor raises this ValueError immediately.

Source

Thrown at src/transformers/feature_extraction_sequence_utils.py:358

        """
        Find the correct padding strategy
        """

        # Get padding strategy
        if padding is not False:
            if padding is True:
                padding_strategy = PaddingStrategy.LONGEST  # Default to pad to the longest sequence in the batch
            elif not isinstance(padding, PaddingStrategy):
                padding_strategy = PaddingStrategy(padding)
            elif isinstance(padding, PaddingStrategy):
                padding_strategy = padding
        else:
            padding_strategy = PaddingStrategy.DO_NOT_PAD

        # Set max length if needed
        if max_length is None:
            if padding_strategy == PaddingStrategy.MAX_LENGTH:
                raise ValueError(
                    f"When setting ``padding={PaddingStrategy.MAX_LENGTH}``, make sure that max_length is defined"
                )

        # Test if we have a padding value
        if padding_strategy != PaddingStrategy.DO_NOT_PAD and (self.padding_value is None):
            raise ValueError(
                "Asking to pad but the feature_extractor does not have a padding value. Please select a value to use"
                " as `padding_value`. For example: `feature_extractor.padding_value = 0.0`."
            )

        return padding_strategy

    def fetch_audio(self, audio_url_or_urls: str | list[str] | list[list[str]], sampling_rate: int | None = None):
        """
        Convert a single or a list of urls into the corresponding `np.ndarray` objects.

        If a single url is passed, the return value will be a single object. If a list is passed a list of objects is
        returned.

View on GitHub (pinned to a597f97485)

Solutions

1. Pass max_length explicitly: fe(audio, padding='max_length', max_length=32000)
2. Or use padding='longest'/padding=True if a fixed length is not required
3. Set a default max_length on the extractor config if it should always apply

Example fix

# before
fe(audio, padding="max_length")

# after
fe(audio, padding="max_length", max_length=32000)

if padding == "max_length" and max_length is None:
    raise ValueError("padding='max_length' needs max_length")  # fail early at the call site
fe(audio, padding=padding, max_length=max_length)

try:
    fe(audio, padding="max_length")
except ValueError as e:
    if "max_length" in str(e):
        fe(audio, padding="longest")  # or supply max_length
    else:
        raise

When it happens

Trigger: Calling fe(audio, padding='max_length') without max_length; also padding=True where the extractor's default strategy was configured as MAX_LENGTH but no length is set anywhere.

Common situations: Assuming max_length defaults from the model config (it does not for feature extractors); migrating tokenizer call patterns verbatim; hand-edited preprocessor configs.

Related errors

• You should supply an instance of `transformers.BatchFeature`
• type of {first_element} unknown: {type(first_element)}. Shou
• Some items in the output dictionary have a different batch s
• Invalid padding strategy:{padding_side}
• When setting ``truncation=True``, make sure that ``max_lengt

AI-assisted analysis of huggingface/transformers@a597f97485 (2026-08-14). Data as JSON: /api/errors/5430dad2518b63b6. Report an issue: GitHub.