Created by: jeng1220

• PaddlePaddle version: 1.5.2
• CPU: Intel i7
• GPU: NVIDIA V100, CUDA 10, cuDNN 7.5.0, TensorRT 5.1 GA
• OS Platform: hub.baidubce.com/paddlepaddle/paddle:1.5.2-gpu-cuda10.0-cudnn7 (the docker image was downloaded from here)

$ nvidia-docker pull hub.baidubce.com/paddlepaddle/paddle:1.5.2-gpu-cuda10.0-cudnn7

• Python version: 2.7

To Reproduce Follow the instructions provided by PaddlePaddle official website to download the model and run the code - resnet50_infer.py:

$ python resnet50_infer.py --model_file ./model/model --params_file ./model/params

All needed files can be found at here

Describe current behavior It threw:

Traceback (most recent call last):
  File "./resnet50_infer.py", line 51, in <module>
    main()    
  File "./resnet50_infer.py", line 22, in main
    inputs = fake_input(args.batch_size)
  File "./resnet50_infer.py", line 39, in fake_input
    image = PaddleTensor(data)
TypeError: __init__(): incompatible constructor arguments. The following argument types are supported:
    1. paddle.fluid.core_avx.PaddleTensor()

If replaced config.disable_gpu() with config.enable_use_gpu(100, 0), then it threw same error message either and NO GPU kernel was invoked (used nsight-system to check it).

Code to reproduce the issue

import argparse                                                                                                                                                                                             
import numpy as np

from paddle.fluid.core import PaddleTensor
from paddle.fluid.core import AnalysisConfig
from paddle.fluid.core import create_paddle_predictor

def main():
    args = parse_args()

    # set AnalysisConfig
    config = AnalysisConfig(args.model_file, args.params_file)
    #config.disable_gpu()
    config.enable_use_gpu(100, 0)

    # create PaddlePredictor
    predictor = create_paddle_predictor(config)

    # set input
    inputs = fake_input(args.batch_size)

    # inference
    outputs = predictor.run(inputs)
    output_num = 512 

    # get output
    output = outputs[0]
    print(output.name)
    output_data = output.as_ndarray() #return numpy.ndarray
    assert list(output_data.shape) == [args.batch_size, output_num]
    for i in range(args.batch_size):
        print(np.argmax(output_data[i]))

def fake_input(batch_size):    
    shape = [batch_size, 3, 318, 318]
    data = np.random.randn(*shape).astype(np.float32)
    image = PaddleTensor(data)
    return [image]

def parse_args():
    parser = argparse.ArgumentParser()
    parser.add_argument("--model_file", type=str, default="model/model", help="model filename")
    parser.add_argument("--params_file", type=str, default="model/params", help="parameter filename")
    parser.add_argument("--batch_size", type=int, default=1, help="batch size")

    return parser.parse_args()

if __name__ == "__main__":
    main()

Other info / logs