huggingface/transformers · error · ValueError

Make sure that all the required parameters: {list(function_a

Error message

Make sure that all the required parameters: {list(function_args.keys())} for {processor.__class__} are passed to the logits processor.

What it means

LogitsProcessorList.__call__ inspects each processor's signature: if __call__ takes more than input_ids/scores, every extra parameter must be present in kwargs. This error fires when a processor needing extra arguments is invoked through a path that did not supply them (e.g. manual list invocation instead of full generate).

Source

Thrown at src/transformers/generation/logits_process.py:90

        Args:
            input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`):
                Indices of input sequence tokens in the vocabulary. [What are input IDs?](../glossary#input-ids)
            scores (`torch.FloatTensor` of shape `(batch_size, config.vocab_size)`):
                Prediction scores of a language modeling head. These can be logits for each vocabulary when not using
                beam search or log softmax for each vocabulary token when using beam search
            kwargs (`dict[str, Any]`, *optional*):
                Additional kwargs that are specific to a logits processor.

        Return:
            `torch.FloatTensor` of shape `(batch_size, config.vocab_size)`:
                The processed prediction scores.

        """
        for processor in self:
            function_args = inspect.signature(processor.__call__).parameters
            if len(function_args) > 2:
                if not all(arg in kwargs for arg in list(function_args.keys())[2:]):
                    raise ValueError(
                        f"Make sure that all the required parameters: {list(function_args.keys())} for "
                        f"{processor.__class__} are passed to the logits processor."
                    )
                scores = processor(input_ids, scores, **kwargs)
            else:
                scores = processor(input_ids, scores)

        return scores


class MinLengthLogitsProcessor(LogitsProcessor):
    r"""
    [`LogitsProcessor`] enforcing a min-length by setting EOS probability to 0. Note that, for decoder-only models
    like most LLMs, the length includes the prompt.

    Args:
        min_length (`int`):
            The minimum length below which the score of `eos_token_id` is set to `-float("Inf")`.

View on GitHub (pinned to a597f97485)

Solutions

1. Pass the missing kwargs: processors(input_ids, scores, **{'attention_mask': am, ...}) — the error lists the required names
2. Or route through model.generate, which supplies all standard kwargs
3. In custom processors, give extra params defaults so len(signature)<=2 logic or kwargs both work

Example fix

# before
scores = processors(input_ids, scores)  # processor needs attention_mask

# after
scores = processors(input_ids, scores, attention_mask=attention_mask)

import inspect
required = [p for proc in processors for p in list(inspect.signature(proc.__call__).parameters)[2:]]
missing = [p for p in set(required) if p not in my_kwargs]
assert not missing, f'missing processor kwargs: {missing}'
scores = processors(input_ids, scores, **my_kwargs)

try:
    scores = processors(input_ids, scores, **kwargs)
except ValueError as e:
    if 'passed to the logits processor' in str(e):
        kwargs.setdefault('attention_mask', attention_mask)  # add commonly missing kwarg
        scores = processors(input_ids, scores, **kwargs)
    else:
        raise

When it happens

Trigger: processors(input_ids, scores) called manually while the list contains e.g. SuppressTokensLogitsProcessor-like processors with extra params, or a custom processor with __call__(self, input_ids, scores, attention_mask) invoked without attention_mask in kwargs.

Common situations: Reusing a LogitsProcessorList built for model.generate in custom decoding loops; processors added by stopping-criteria machinery that expect generate-managed kwargs; signature changes across versions adding new params.

Related errors

• {self.__class__} is an abstract class. Only classes inheriti
• Cannot assign to field {name}, you should create a new insta
• Framework '{return_tensors}' not recognized!
• return_tensors should be `'pt'` or `None`
• Can only set a dictionary as `tp_plan`

AI-assisted analysis of huggingface/transformers@a597f97485 (2026-08-14). Data as JSON: /api/errors/233481b4ec45443f. Report an issue: GitHub.