[TRT] UffParser: Output error: Output GPU_0/tower_0/Softmax not found

hi, what is the new output_name for resnet50-infer-5.uff?. This command used to work

./trtexec --uff=/workspace/tensorrt/data/resnet50/resnet50-infer-5.uff --batch=1 --output=GPU_0/tower_0/Softmax --uffInput=input,3,224,224

But now I am getting the error:

[TRT] UffParser: Output error: Output GPU_0/tower_0/Softmax not found

Hi,

Could you please provide details on the platforms you are using:
o Linux distro and version
o GPU type
o Nvidia driver version
o CUDA version
o CUDNN version
o Python version [if using python]
o Tensorflow and PyTorch version
o TensorRT version
If possible, please share the model file to reproduce the issue.

Thanks

Hi SunilJB, here is the requested information:
o Linux distro and version: Ubuntu 18.04.4 LTS
o GPU type: RTX8000
o Nvidia driver version: 435.21
o CUDA version: 10.1
o CUDNN version: 7.6.4
o Python version [if using python]: 3.6
o Tensorflow version: n/a
o TensorRT version: TensorRT 6.0.1

If possible, please share the model file to reproduce the issue:
Running trtexec via docker image: nvcr.io/nvidia/tensorrt:19.10-py3. See resnet50-infer-5.uff model located within the docker at /workspace/tensorrt/data/resnet50/

I need to know what is the new output_node name of the graph, it used to be GPU_0/tower_0/Softmax

Hi,

Please try below command, it seems to be working on TRT 7 NGC container:

trtexec --uff=resnet50-infer-5.uff --batch=1 --output=output --uffInput=input,3,224,224 --verbose

Even in python example following model details are used:

MODEL_PATH = "resnet50-infer-5.uff"
    INPUT_NAME = "input"
    INPUT_SHAPE = (3, 224, 224)
    OUTPUT_NAME = "output"

Thanks