PaddlePaddle/PaddleOCR · error · NotImplementedError

the datatype is not supported

Error message

the datatype is not supported

What it means

NotImplementedError from the CLI of ppocr/utils/formula_utils/unimernet_data_convert.py: the --datatype argument only accepts exactly 'unimernet_train' or 'unimernet_test'; any other string falls through both branches and raises.

Source

Thrown at ppocr/utils/formula_utils/unimernet_data_convert.py:107

        "--output_path", type=str, default="out_label.txt", help="Output file name"
    )
    parser.add_argument(
        "--datatype", type=str, default="out_label.txt", help="datatype"
    )
    args = parser.parse_args()
    if args.datatype == "unimernet_train":
        latexocr2paddleocr_train(
            args.image_dir,
            args.unimernet_txt_path,
            args.hme100k_txt_path,
            args.output_path,
        )
    elif args.datatype == "unimernet_test":
        unimernet2paddleocr_test(
            args.image_dir, args.unimernet_txt_path, args.output_path
        )
    else:
        raise NotImplementedError("the datatype is not supported")

View on GitHub (pinned to 2661c7c0ef)

Solutions

1. Use exactly --datatype unimernet_train or --datatype unimernet_test.
2. Check the script's argument help/source for the supported values before running.
3. For a validation split, convert it with unimernet_test if that matches your evaluation pipeline.

Example fix

# before
python unimernet_data_convert.py --datatype unimernet_val ...

# after
python unimernet_data_convert.py --datatype unimernet_test ...

ALLOWED = {"unimernet_train", "unimernet_test"}
if args.datatype not in ALLOWED:
    raise SystemExit(f"--datatype must be one of {sorted(ALLOWED)}, got {args.datatype!r}")

def is_supported_datatype(v) -> bool:
    return v in {"unimernet_train", "unimernet_test"}

When it happens

Trigger: Running the UniMERNet data conversion script with a typo'd or unsupported --datatype, e.g. 'unimernet_val', 'train', or 'Unimernet_Test'.

Common situations: Preparing LaTeX/formula recognition training data; users assume 'val'/'test' split vocabulary from other conversion scripts; case or underscore mistakes when copying the command from docs.

Related errors

• Could not parse xcresulttool JSON: {e}.
• xcresulttool export attachments failed with exit {r.returnco
• Could not resolve exported file (wanted {wanted_out_name!r},
• xcresulttool metrics export failed
• error: {msg}

AI-assisted analysis of PaddlePaddle/PaddleOCR@2661c7c0ef (2026-08-14). Data as JSON: /api/errors/5c745b99a57ac200. Report an issue: GitHub.