PaddlePaddle/PaddleOCR · 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 __call__ signature; any parameter beyond (input_ids, scores) must be supplied via kwargs. If one is missing, it raises this ValueError because calling the processor would itself raise a TypeError, and the explicit message names the expected parameters.

Source

Thrown at ppocr/modeling/heads/rec_unimernet_head.py:1569

        )

        return attn_output, attn_output_weights


class LogitsProcessorList(list):
    """
    A list of logits processors that can be applied sequentially.

    Methods:
        __call__(input_ids, scores, **kwargs): Apply all processors to the given inputs.
    """

    def __call__(self, input_ids, scores, **kwargs):
        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 ForcedEOSTokenLogitsProcessor(object):
    """
    A processor that forces the generation of an end-of-sequence (EOS) token
    at a specified position in the sequence.

    This is typically used in language generation tasks to ensure that the
    generated sequence ends properly when it reaches a certain length.

    Args:

View on GitHub (pinned to 2661c7c0ef)

Solutions

1. Pass the missing kwarg(s) into the logits_processor call, e.g. logits_processor(input_ids, scores, min_length=cfg.min_length, eos_token_id=cfg.eos_token_id)
2. Or remove the processor that requires unavailable arguments
3. Wrap processor construction so required args are bound via functools.partial, reducing the signature to (input_ids, scores)

Example fix

# before
next_tokens_scores = self.logits_processor(input_ids, next_token_logits)
# after
next_tokens_scores = self.logits_processor(
    input_ids, next_token_logits,
    min_length=self.config.min_length,
    eos_token_id=eos_token_id,
)

import inspect
required = [p for p in inspect.signature(processor.__call__).parameters][2:]
missing = [a for a in required if a not in generation_kwargs]
assert not missing, f'logits processor missing kwargs: {missing}'

try:
    scores = logits_processor(input_ids, scores, **gen_kwargs)
except ValueError as e:
    if 'passed to the logits processor' in str(e):
        gen_kwargs.update(min_length=cfg.min_length, eos_token_id=eos_id)
        scores = logits_processor(input_ids, scores, **gen_kwargs)
    else:
        raise

When it happens

Trigger: Running generation with processors such as MinLengthLogitsProcessor (needs min_length) or ForcedEOSTokenLogitsProcessor (needs forced eos id) while the generate loop / custom loop does not pass those kwargs to logits_processor(input_ids, scores, **kwargs).

Common situations: Custom generate implementations (like the one in this head) that forward a fixed set of kwargs; adding a new processor to logits_processor list without updating the generation loop to supply its arguments.

Related errors

• `decoder_start_token_id` or `bos_token_id` has to be defined
• You have to specify either decoder_input_ids or decoder_inpu
• `decoder_start_token_id` expected to have length {batch_size
• If `eos_token_id` is defined, make sure that `pad_token_id`
• Please set inference model dir in Global.inference_model or

AI-assisted analysis of PaddlePaddle/PaddleOCR@2661c7c0ef (2026-08-14). Data as JSON: /api/errors/9410a28c75895dc2. Report an issue: GitHub.