sgl-project/sglang · error · ValueError

At least one of text, input_ids, or image should be provided

Error message

At least one of text, input_ids, or image should be provided

What it means

normalize_batch_and_arguments requires at least one input modality: text, input_ids, or image_data. All three being None means there is nothing to tokenize/embed, so the request is rejected up front.

Source

Thrown at python/sglang/srt/managers/io_struct.py:1180

        if isinstance(self.rid, list):
            self.rid = [uuid.uuid4().hex for _ in range(len(self.rid))]
        else:
            self.rid = uuid.uuid4().hex
        return self.rid

    def _validate_rid_uniqueness(self):
        """Validate that request IDs within a batch are unique."""
        if isinstance(self.rid, list) and len(set(self.rid)) != len(self.rid):
            counts = Counter(self.rid)
            duplicates = [rid for rid, count in counts.items() if count > 1]
            raise ValueError(
                f"Duplicate request IDs detected within the request: {duplicates}"
            )

    def normalize_batch_and_arguments(self):
        # at least one of text, input_ids, or image should be provided
        if self.text is None and self.input_ids is None and self.image_data is None:
            raise ValueError(
                "At least one of text, input_ids, or image should be provided"
            )

        # text and input_ids cannot be provided at the same time
        if self.text is not None and self.input_ids is not None:
            raise ValueError("text and input_ids cannot be provided at the same time")

        # Derive the batch size
        self.batch_size = 0
        self.is_single = True

        # check the batch size of text
        if self.text is not None:
            if isinstance(self.text, list):
                self.batch_size += len(self.text)
                self.is_single = False
            else:
                self.batch_size += 1

View on GitHub (pinned to 0132848349)

Solutions

1. Supply text='...', input_ids=[...], or image_data=[...]
2. Debug why the prompt field is None before constructing the request
3. For embed-only or multimodal calls, still include the text or image

Example fix

// before
req = GenerateReqInput(sampling_params={'temperature':0.7})
// after
req = GenerateReqInput(text='Hello', sampling_params={'temperature':0.7})

assert text is not None or input_ids is not None or image_data is not None

When it happens

Trigger: GenerateReqInput(sampling_params={...}) with no text/input_ids/image_data; a client sends JSON {} or only sampling fields; passing empty string via a variable that is actually None.

Common situations: Client bug where the prompt variable is None due to upstream extraction failure; building requests programmatically from possibly-empty records.

Related errors

• Unknown req_type: {req_type!r} (expected 'generate' or 'embe
• text and input_ids cannot be provided at the same time
• token_ids_logprob must be a flat list of integers.
• Unknown serve backend {name!r}. Available values: {available
• Multiple distributions register serve backend {name!r}: {pro

AI-assisted analysis of sgl-project/sglang@0132848349 (2026-08-28). Data as JSON: /api/errors/5bc0b7a3dca5e385. Report an issue: GitHub.