sgl-project/sglang · error · ValueError

lora_nickname cannot be empty

Error message

lora_nickname cannot be empty

What it means

ValueError guard in SGLDiffusionServerAPI.set_lora: lora_nickname is empty/None. The LoRA hot-swap endpoint requires a nickname identifying which trained adapter to activate.

Source

Thrown at python/sglang/multimodal_gen/apps/ComfyUI_SGLDiffusion/core/server_api.py:474

    ) -> Dict[str, Any]:
        """
        Set a LoRA adapter for the specified transformer(s).

        Args:
            lora_nickname: The nickname of the adapter (required).
            lora_path: Path to the LoRA adapter (local path or HF repo id).
                      Required for the first load; optional if re-activating a cached nickname.
            target: Which transformer(s) to apply the LoRA to. One of:
                - "all": Apply to all transformers (default)
                - "transformer": Apply only to the primary transformer (high noise for Wan2.2)
                - "transformer_2": Apply only to transformer_2 (low noise for Wan2.2)
                - "critic": Apply only to the critic model

        Returns:
            Dictionary containing the API response with status and message
        """
        if not lora_nickname:
            raise ValueError("lora_nickname cannot be empty")

        # Prepare request payload
        payload: Dict[str, Any] = {
            "lora_nickname": lora_nickname,
            "target": target,
        }

        # Add optional lora_path if provided
        if lora_path:
            payload["lora_path"] = lora_path

        try:
            response = requests.post(
                f"{self.base_url}/set_lora",
                json=payload,
                headers=self.headers,
                timeout=30,
            )

View on GitHub (pinned to 0132848349)

Solutions

1. Pass the exact nickname the adapter was registered under on the server.
2. Fetch the server's LoRA list first and validate the nickname against it.
3. Default empty inputs to a known adapter or skip the call entirely.

Example fix

// before
api.set_lora("")
// after
if lora_nickname:
    api.set_lora(lora_nickname)
else:
    print("no LoRA selected; skipping")

loras = api.list_loras() if hasattr(api, 'list_loras') else None
if loras is not None:
    assert lora_nickname in {l['nickname'] for l in loras}

def valid_nickname(name: str | None) -> bool:
    return isinstance(name, str) and bool(name.strip())

When it happens

Trigger: Calling set_lora(lora_nickname="") — often because a variable holding the nickname was never set or the ComfyUI dropdown left empty.

Common situations: Nickname typo'd as empty string; LoRA list not loaded before the call; unconfigured optional input passed through as None.

Related errors

• LoRA batch_info must provide max_len or seg_lens.
• No image data in response
• Image index {index} out of range
• No base64 image data found
• Failed to set LoRA adapter: {str(e)}

AI-assisted analysis of sgl-project/sglang@0132848349 (2026-08-28). Data as JSON: /api/errors/9555c53e22f5d224. Report an issue: GitHub.