Comfy-Org/ComfyUI · error · KeyError

Expected key `target` to instantiate.

Error message

Expected key `target` to instantiate.

What it means

instantiate_from_config() builds an object from a config dict by importing the class named under the "target" key. The only exempt values are the sentinel strings '__is_first_stage__' and '__is_unconditional__' (which return None); any other dict without "target" raises KeyError. This is a config-schema error: the dict cannot describe what to instantiate.

Source

Thrown at comfy/ldm/util.py:79

    Take the mean over all non-batch dimensions.
    """
    return tensor.mean(dim=list(range(1, len(tensor.shape))))


def count_params(model, verbose=False):
    total_params = sum(p.numel() for p in model.parameters())
    if verbose:
        logging.info(f"{model.__class__.__name__} has {total_params*1.e-6:.2f} M params.")
    return total_params


def instantiate_from_config(config):
    if "target" not in config:
        if config == '__is_first_stage__':
            return None
        elif config == "__is_unconditional__":
            return None
        raise KeyError("Expected key `target` to instantiate.")
    return get_obj_from_str(config["target"])(**config.get("params", dict()))


def get_obj_from_str(string, reload=False):
    module, cls = string.rsplit(".", 1)
    if reload:
        module_imp = importlib.import_module(module)
        importlib.reload(module_imp)
    return getattr(importlib.import_module(module, package=None), cls)


class AdamWwithEMAandWings(optim.Optimizer):
    # credit to https://gist.github.com/crowsonkb/65f7265353f403714fce3b2595e0b298
    def __init__(self, params, lr=1.e-3, betas=(0.9, 0.999), eps=1.e-8,  # TODO: check hyperparameters before using
                 weight_decay=1.e-2, amsgrad=False, ema_decay=0.9999,   # ema decay to match previous code
                 ema_power=1., param_names=()):
        """AdamW that saves EMA versions of the parameters."""
        if not 0.0 <= lr:

View on GitHub (pinned to 1c6d8d45b3)

Solutions

  1. Ensure the dict has a "target" key with a fully qualified class path, e.g. {"target": "comfy.ldm.seedvr.vae.AutoencoderKLSeedVR", "params": {...}}.
  2. If None is intended for an optional stage, use the sentinels '__is_first_stage__' or '__is_unconditional__' or skip the call.
  3. Inspect the checkpoint's config file for a truncated/malformed stage entry.

Example fix

# before
model = instantiate_from_config({"params": {"embed_dim": 4}})
# after
model = instantiate_from_config({"target": "comfy.ldm....Decoder", "params": {"embed_dim": 4}})
Defensive patterns

Strategy: validation

Validate before calling

SENTINELS = {"__is_first_stage__", "__is_unconditional__"}
if isinstance(config, dict) and "target" not in config and config not in SENTINELS:
    raise KeyError("config lacks 'target'; wrap as {'target': ..., 'params': ...}")
obj = instantiate_from_config(config)

Type guard

def is_instantiable_config(config) -> bool:
    return config in ("__is_first_stage__", "__is_unconditional__") or (isinstance(config, dict) and "target" in config)

Try / catch

try:
    obj = instantiate_from_config(stage_cfg)
except KeyError as e:
    if "target" in str(e):
        logging.warning("skipping stage without target: %r", stage_cfg)
        obj = None
    else:
        raise

Prevention

When it happens

Trigger: instantiate_from_config(config) where config is a dict lacking "target" — e.g. a checkpoint's JSON/YAML config missing the first-stage/unet entry, or passing params directly instead of wrapping them in {"target": ..., "params": ...}.

Common situations: Editing or truncating a model config JSON; a checkpoint config that uses a different schema version; passing the wrong sub-dict (config["params"] instead of config).

Related errors

AI-assisted analysis of Comfy-Org/ComfyUI@1c6d8d45b3 (2026-08-14). Data as JSON: /api/errors/bdcebd9a2ee60773. Report an issue: GitHub.