invoke-ai/InvokeAI · warning · NotAMatchError

model looks like an Anima LoRA, not a Stable Diffusion LoRA

Error message

model looks like an Anima LoRA, not a Stable Diffusion LoRA

What it means

When detecting the base of a Stable Diffusion LoRA, `_get_base_or_raise` first rules out Anima LoRAs: their `lora_te_` text-encoder key shapes would make `lora_token_vector_length()` misreport the base as SD2/SDXL. If Cosmos-DiT Kohya or PEFT key patterns are found, the model is identified as an Anima LoRA and `NotAMatchError` is raised so the SD LoRA config classes reject it.

Source

Thrown at invokeai/backend/model_manager/configs/lora.py:651

        )

        if not has_key_with_lora_prefix and not has_key_with_lora_suffix:
            raise NotAMatchError("model does not match LyCORIS LoRA heuristics")

    @classmethod
    def _get_base_or_raise(cls, mod: ModelOnDisk) -> BaseModelType:
        if _get_flux_lora_format(mod):
            if _is_flux2_lora(mod):
                return BaseModelType.Flux2
            return BaseModelType.Flux

        state_dict = mod.load_state_dict()
        str_keys = [k for k in state_dict.keys() if isinstance(k, str)]

        # Rule out Anima LoRAs — their lora_te_ keys have shapes that
        # lora_token_vector_length() misidentifies as SD2/SDXL.
        if has_cosmos_dit_kohya_keys(str_keys) or has_cosmos_dit_peft_keys(str_keys):
            raise NotAMatchError("model looks like an Anima LoRA, not a Stable Diffusion LoRA")

        # If we've gotten here, we assume that the model is a Stable Diffusion model
        token_vector_length = lora_token_vector_length(state_dict)
        if token_vector_length == 768:
            return BaseModelType.StableDiffusion1
        elif token_vector_length == 1024:
            return BaseModelType.StableDiffusion2
        elif token_vector_length == 1280:
            return BaseModelType.StableDiffusionXL  # recognizes format at https://civitai.com/models/224641
        elif token_vector_length == 2048:
            return BaseModelType.StableDiffusionXL
        # Some SDXL LoRAs (e.g. self-attention-only "slider" LoRAs) target only the UNet
        # and lack the cross-attention / text-encoder keys that lora_token_vector_length()
        # needs. Fall back to detecting SDXL from the UNet's deep transformer-block structure.
        elif _state_dict_looks_like_sdxl_unet_lora(state_dict):
            return BaseModelType.StableDiffusionXL
        else:
            raise NotAMatchError(f"unrecognized token vector length {token_vector_length}")

View on GitHub (pinned to 0b6a024f2f)

Solutions

1. Confirm the LoRA's source model — if it's Anima, use it only with Anima checkpoints / a config that supports it
2. Re-export or convert the Anima LoRA to standard SD/SDXL LoRA key format if it was mislabeled
3. Remove the Anima LoRA from the SD LoRA import path and register it under the correct model type
4. Upgrade InvokeAI to a version with dedicated Anima LoRA support

Example fix

// before: importing an Anima (Cosmos-DiT keys) file as SD LoRA
SD_LoRA_Config.from_model_on_disk(anima_lora_dir)  # NotAMatchError
// after: use a config/matcher for Anima, or convert keys to lora_te_/lora_unet_ SD format
Anima_LoRA_Config.from_model_on_disk(anima_lora_dir)

from invokeai.backend.model_manager.util.lora_conversions import has_cosmos_dit_kohya_keys, has_cosmos_dit_peft_keys
sd = mod.load_state_dict()
keys = [k for k in sd.keys() if isinstance(k, str)]
if has_cosmos_dit_kohya_keys(keys) or has_cosmos_dit_peft_keys(keys):
    print("Anima LoRA — do not import as SD LoRA")

def is_anima_lora(keys: list[str]) -> bool:
    from invokeai.backend.model_manager.util.lora_conversions import has_cosmos_dit_kohya_keys, has_cosmos_dit_peft_keys
    return has_cosmos_dit_kohya_keys(keys) or has_cosmos_dit_peft_keys(keys)

try:
    cfg = StableDiffusionLoRAConfig.from_model_on_disk(mod)
except NotAMatchError as e:
    if "Anima" in str(e):
        print("use Anima-compatible tooling or convert the LoRA")

When it happens

Trigger: `from_model_on_disk` on an SD-family LoRA probe where the state dict contains `has_cosmos_dit_kohya_keys` or `has_cosmos_dit_peft_keys` patterns (Anima LoRA weights) — i.e. importing an Anima LoRA through the Stable Diffusion LoRA config path.

Common situations: Autoimport folder containing Anima LoRAs alongside SD LoRAs; downloading an Anima LoRA believing it is SD1.5/SDXL-compatible; a converted Anima LoRA retaining Cosmos-DiT key naming.

Related errors

• model does not match Anima LoRA heuristics
• Unsupported control_lllite type: {type(control_lllite)}
• The Anima ControlNet-LLLite model '{lllite_field.control_mod
• This Anima ControlNet-LLLite adapter is an inpainting adapte
• Unsupported Anima ControlNet-LLLite adapter: expected 3 or 4

AI-assisted analysis of invoke-ai/InvokeAI@0b6a024f2f (2026-08-29). Data as JSON: /api/errors/40dcd9a01f025f76. Report an issue: GitHub.