hankcs/HanLP · error · ValueError

You cannot specify both input_ids and inputs_embeds at the s

Error message

You cannot specify both input_ids and inputs_embeds at the same time

What it means

CategoricalAccuracy supports tie_break (handling multiple classes sharing the max predicted score as correct) only when scoring the single top prediction (top_k=1). Enabling tie_break with top_k > 1 is a contradictory configuration and raises ValueError at construction.

Source

Thrown at hanlp/components/amr/amrbart/model_interface/modeling_bart.py:789

                than the model's internal embedding lookup matrix.
            output_attentions (`bool`, *optional*):
                Whether or not to return the attentions tensors of all attention layers. See `attentions` under
                returned tensors for more detail.
            output_hidden_states (`bool`, *optional*):
                Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors
                for more detail.
            return_dict (`bool`, *optional*):
                Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple.
        """
        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
        output_hidden_states = (
            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
        )
        return_dict = return_dict if return_dict is not None else self.config.use_return_dict

        # retrieve input_ids and inputs_embeds
        if input_ids is not None and inputs_embeds is not None:
            raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")
        elif input_ids is not None:
            input_shape = input_ids.size()
            input_ids = input_ids.view(-1, input_shape[-1])
        elif inputs_embeds is not None:
            input_shape = inputs_embeds.size()[:-1]
        else:
            raise ValueError("You have to specify either input_ids or inputs_embeds")

        if inputs_embeds is None:
            inputs_embeds = self.embed_tokens(input_ids) * self.embed_scale

        embed_pos = self.embed_positions(input_shape)

        hidden_states = inputs_embeds + embed_pos
        hidden_states = self.layernorm_embedding(hidden_states)
        hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training)

        # expand attention_mask

View on GitHub (pinned to ddb1299bdd)

Solutions

1. Set top_k=1 when tie_break=True
2. Disable tie_break if you need top_k > 1 scoring
3. Use a different metric for tie-aware top-k evaluation

Example fix

# before
metric = CategoricalAccuracy(top_k=5, tie_break=True)
# after
metric = CategoricalAccuracy(top_k=5, tie_break=False)

assert not (top_k > 1 and tie_break), 'tie_break requires top_k == 1'

try:
    m = CategoricalAccuracy(top_k=top_k, tie_break=tie_break)
except ValueError:
    m = CategoricalAccuracy(top_k=top_k, tie_break=False)

When it happens

Trigger: Constructing CategoricalAccuracy(top_k=5, tie_break=True).

Common situations: Copy-pasting metric configs and toggling both flags; enabling tie_break to fix ambiguous predictions while leaving top_k from a previous top-k experiment.

Understand the failure class

Background: Config validation failed: what "invalid value for {key}" and settings-rejection errors mean across 19 open-source libraries — this error's family across 19 libraries.

Related errors

• You have to specify either input_ids or inputs_embeds
• Unsupported argument type: {item}
• Unrecognized mapper type {mapper}
• self.model.config.pad_token_id has to be defined.
• embed_dim must be divisible by num_heads (got `embed_dim`: {

AI-assisted analysis of hankcs/HanLP@ddb1299bdd (2026-08-27). Data as JSON: /api/errors/289fb0dda3619279. Report an issue: GitHub.