invoke-ai/InvokeAI · warning · NotAMatchError

base is {recognized_base}, not {expected_base}

Error message

base is {recognized_base}, not {expected_base}

What it means

`NotAMatchError` from `T2IAdapterDiffusersConfig._validate_base`: the adapter's `adapter_type` (read from its diffusers config.json) resolved to a base model that does not match the `base` literal declared by this specific config class. It is part of normal probing — each SD1/SDXL config class validates the resolved base and declines non-matching adapters so the right class (e.g. `T2IAdapter_Diffusers_SD1_Config` vs the XL variant) picks the model up.

Source

Thrown at invokeai/backend/model_manager/configs/t2i_adapter.py:57

        raise_for_class_name(
            common_config_paths(mod.path),
            {
                "T2IAdapter",
            },
        )

        cls._validate_base(mod)

        return cls(**override_fields)

    @classmethod
    def _validate_base(cls, mod: ModelOnDisk) -> None:
        """Raise `NotAMatch` if the model base does not match this config class."""
        expected_base = cls.model_fields["base"].default
        recognized_base = cls._get_base_or_raise(mod)
        if expected_base is not recognized_base:
            raise NotAMatchError(f"base is {recognized_base}, not {expected_base}")

    @classmethod
    def _get_base_or_raise(cls, mod: ModelOnDisk) -> BaseModelType:
        config_dict = get_config_dict_or_raise(common_config_paths(mod.path))

        adapter_type = config_dict.get("adapter_type")

        match adapter_type:
            case "full_adapter_xl":
                return BaseModelType.StableDiffusionXL
            case "full_adapter" | "light_adapter":
                return BaseModelType.StableDiffusion1
            case _:
                raise NotAMatchError(f"unrecognized adapter_type '{adapter_type}'")


class T2IAdapter_Diffusers_SD1_Config(T2IAdapter_Diffusers_Config_Base, Config_Base):
    base: Literal[BaseModelType.StableDiffusion1] = Field(default=BaseModelType.StableDiffusion1)

View on GitHub (pinned to 0b6a024f2f)

Solutions

1. No action needed during auto-import — the matching base's config class will accept the model
2. If you forced a specific config class, switch to the one whose `base` matches `adapter_type` (`full_adapter`/`light_adapter` → SD1, `full_adapter_xl` → SDXL)
3. Fix a wrong `adapter_type` in the adapter's config.json if the checkpoint was mislabeled during conversion

Example fix

// before: registering full_adapter_xl with T2IAdapter_Diffusers_SD1_Config
// after: use the SDXL config class (base=StableDiffusionXL) or fix config.json adapter_type

import json
from pathlib import Path

def adapter_base(model_dir: Path) -> str | None:
    cfg = json.loads(next(model_dir.glob("**/config.json")).read_text())
    t = cfg.get("adapter_type")
    return {"full_adapter_xl": "sdxl", "full_adapter": "sd1", "light_adapter": "sd1"}.get(t)

try:
    cfg = T2IAdapter_Diffusers_SD1_Config.from_model_on_disk(mod, override_fields)
except NotAMatchError as e:
    if str(e).startswith("base is"):
        print("Adapter belongs to a different base — let the prober pick the right class")
    raise

When it happens

Trigger: `from_model_on_disk` on a diffusers T2I-Adapter folder whose `config.json` `adapter_type` maps to a different `BaseModelType` than the class being probed (e.g. `full_adapter_xl` probed by the SD1 config class).

Common situations: Auto-import scanning where the SD1 class sees an XL adapter (expected; the XL class will match instead), manually registering a model with an explicit config class that disagrees with `adapter_type`, adapters converted with a mismatched `adapter_type` field.

Related errors

• Unexpected T2I-Adapter base model type: '{model_config.base}
• Unexpected T2I-Adapter base model type: '${t2i_adapter_model
• model does not look like an Anima LoRA
• base is {recognized_base}, not {expected_base}
• model does not match SpandrelImageToImage heuristics

AI-assisted analysis of invoke-ai/InvokeAI@0b6a024f2f (2026-08-29). Data as JSON: /api/errors/c76094afe07c0c75. Report an issue: GitHub.