ultralytics/yolov5 · error · ValueError

Invalid model path {w}. Provide model directory or a .pdipar

Error message

Invalid model path {w}. Provide model directory or a .pdiparams file.

What it means

In the PaddlePaddle branch of DetectMultiBackend, ValueError is raised when the weight path is neither a directory nor a file with the .pdiparams suffix. The loader expects either the export directory containing model.json + *.pdiparams, or the .pdiparams file itself (it then looks for a sibling model.json). Any other shape (e.g. a bare .json, a .pdmodel, or a random file) hits this branch.

Source

Thrown at models/common.py:654

                meta = ast.literal_eval(model.read(meta_file).decode("utf-8"))
                stride, names = int(meta["stride"]), meta["names"]
        elif tfjs:  # TF.js
            raise NotImplementedError("ERROR: YOLOv5 TF.js inference is not supported")
        # PaddlePaddle
        elif paddle:
            LOGGER.info(f"Loading {w} for PaddlePaddle inference...")
            check_requirements("paddlepaddle-gpu" if cuda else "paddlepaddle>=3.0.0")
            import paddle.inference as pdi

            w = Path(w)
            if w.is_dir():
                model_file = next(w.rglob("*.json"), None)
                params_file = next(w.rglob("*.pdiparams"), None)
            elif w.suffix == ".pdiparams":
                model_file = w.with_name("model.json")
                params_file = w
            else:
                raise ValueError(f"Invalid model path {w}. Provide model directory or a .pdiparams file.")

            if not (model_file and params_file and model_file.is_file() and params_file.is_file()):
                raise FileNotFoundError(f"Model files not found in {w}. Both .json and .pdiparams files are required.")

            config = pdi.Config(str(model_file), str(params_file))
            if cuda:
                config.enable_use_gpu(memory_pool_init_size_mb=2048, device_id=0)
            config.disable_mkldnn()  # disable MKL-DNN for PIR compatibility
            predictor = pdi.create_predictor(config)
            input_handle = predictor.get_input_handle(predictor.get_input_names()[0])
            output_names = predictor.get_output_names()

        elif triton:  # NVIDIA Triton Inference Server
            LOGGER.info(f"Using {w} as Triton Inference Server...")
            check_requirements("tritonclient[all]")
            from utils.triton import TritonRemoteModel

            model = TritonRemoteModel(url=w)

View on GitHub (pinned to 20d1d78a08)

Solutions

1. Point at the export directory produced by export.py --include paddle, which contains model.json and model.pdiparams.
2. Or point directly at the .pdiparams file; the loader pairs it with a sibling model.json automatically.
3. Do not pass .pdmodel or .json paths; they are not accepted inputs.

Example fix

# before
model = DetectMultiBackend('model.json')

# after
model = DetectMultiBackend('yolov5s_paddle_model/')  # dir with model.json + model.pdiparams

from pathlib import Path

def valid_paddle_path(w) -> bool:
    p = Path(w)
    return p.is_dir() or p.suffix == '.pdiparams'

from pathlib import Path

def is_paddle_loadable(w: str) -> bool:
    """True if DetectMultiBackend's paddle branch accepts this path shape."""
    p = Path(w)
    return p.is_dir() or p.suffix == ".pdiparams"

When it happens

Trigger: Passing DetectMultiBackend('model.json') or a .pdmodel file; passing the directory of a Paddle *training* checkpoint rather than a Paddle Inference export; a typo'd path with a wrong extension.

Common situations: Converting models with paddle tools and handing DetectMultiBackend the wrong artifact of the set; copying only some files out of the export directory and renaming them.

Related errors

• Model files not found in {w}. Both .json and .pdiparams file
• ERROR: {w} is not a supported format
• ERROR: YOLOv5 TF.js inference is not supported
• Source path '{source}' does not exist
• TensorRT engine deserialization failed. Re-export the engine

AI-assisted analysis of ultralytics/yolov5@20d1d78a08 (2026-08-15). Data as JSON: /api/errors/71579545726e5c00. Report an issue: GitHub.