AUTOMATIC1111/stable-diffusion-webui · error · Exception

data.pkl not found in {filename}

Error message

data.pkl not found in {filename}

What it means

After passing the zip-filename check, check_pt looks for exactly one member matching '<dir>/data.pkl' — the pickle that describes the stored tensors. Zero matches means the archive is a valid zip but has no PyTorch data.pkl, so it cannot be a new-format torch checkpoint, and the Exception is raised (the subsequent BadZipfile branch for legacy format is not taken because the file opened as a zip fine).

Source

Thrown at modules/safe.py:89

def check_zip_filenames(filename, names):
    for name in names:
        if allowed_zip_names_re.match(name):
            continue

        raise Exception(f"bad file inside {filename}: {name}")


def check_pt(filename, extra_handler):
    try:

        # new pytorch format is a zip file
        with zipfile.ZipFile(filename) as z:
            check_zip_filenames(filename, z.namelist())

            # find filename of data.pkl in zip file: '<directory name>/data.pkl'
            data_pkl_filenames = [f for f in z.namelist() if data_pkl_re.match(f)]
            if len(data_pkl_filenames) == 0:
                raise Exception(f"data.pkl not found in {filename}")
            if len(data_pkl_filenames) > 1:
                raise Exception(f"Multiple data.pkl found in {filename}")
            with z.open(data_pkl_filenames[0]) as file:
                unpickler = RestrictedUnpickler(file)
                unpickler.extra_handler = extra_handler
                unpickler.load()

    except zipfile.BadZipfile:

        # if it's not a zip file, it's an old pytorch format, with five objects written to pickle
        with open(filename, "rb") as file:
            unpickler = RestrictedUnpickler(file)
            unpickler.extra_handler = extra_handler
            for _ in range(5):
                unpickler.load()


def load(filename, *args, **kwargs):

View on GitHub (pinned to 82a973c043)

Solutions

1. Verify the file is a real PyTorch checkpoint: `python -c "import zipfile;print(zipfile.ZipFile('model.ckpt').namelist())"` should show data.pkl and data/* entries.
2. Re-download the model from its official source — a truncated or proxy-replaced download often yields a small invalid zip.
3. Remove non-checkpoint archives from the models directories so the scanner doesn't try to load them.

Example fix

# before
mv photos.zip model.ckpt   # renamed archive -> 'data.pkl not found'

# after: use a genuine torch export
# torch.save(state_dict, 'model.ckpt')  # archive contains data.pkl + data/0...

import re, zipfile

def has_data_pkl(path):
    pat = re.compile(r'^([^/]+)/data\.pkl$')
    try:
        with zipfile.ZipFile(path) as z:
            return any(pat.match(n) for n in z.namelist())
    except zipfile.BadZipFile:
        return False  # legacy (non-zip) torch format is handled separately

assert has_data_pkl('model.ckpt'), 'not a new-format torch checkpoint; check the file type/download'

def looks_like_torch_zip_checkpoint(path: str) -> bool:
    import re, zipfile
    pat = re.compile(r'^([^/]+)/data\.pkl$')
    try:
        with zipfile.ZipFile(path) as z:
            return any(pat.match(n) for n in z.namelist())
    except zipfile.BadZipFile:
        return False

try:
    weights = torch.load(path)
except Exception as e:
    if 'data.pkl not found' in str(e):
        raise RuntimeError(f'{path} is a zip but not a torch checkpoint; verify the download/source')
    raise

When it happens

Trigger: Passing a file to torch.load/webui loading that is a valid zip archive but not a torch checkpoint — e.g. a renamed .zip, a .npz, an ONNX file, or a torch .pt' archive whose data.pkl sits at a nonstandard path already rejected earlier.

Common situations: Renaming unrelated archives to .ckpt/.pt; corrupted downloads that are actually HTML/error zips; model files from other frameworks accidentally placed in models/Stable-diffusion.

Related errors

• global '{module}/{name}' is forbidden
• bad file inside {filename}: {name}
• Requests not allowed
• Request to local resource not allowed
• When merging inpainting model with a normal one, A must be t

AI-assisted analysis of AUTOMATIC1111/stable-diffusion-webui@82a973c043 (2026-08-14). Data as JSON: /api/errors/f455595de182c0fe. Report an issue: GitHub.