Need help authoring Model configuration for Pytorch MNIST

Hi I am trying to host my first model in triton.

• Ran the MNIST model from examples/main.py at https://github.com/pytorch/examples/tree/master/mnist
• Saved model to “mnist_cnn.pt”
• Converted saved model to torchscript format.
• Created the following config and followed the steps at https://github.com/triton-inference-server/server/blob/main/docs/model_configuration.md to create a model repository

When I to run inference I get error saying mnist model failed, bad request.

platform: "pytorch_libtorch"
max_batch_size: 0
input [
  {
    name: "INPUT__0"
    data_type: TYPE_FP32
    dims: [ 28,28 ]
  }
]
output [
  {
    name: "output__0"
    data_type: TYPE_FP32
    dims: [ 10 ]
  }
]

What am I missing?

[CoderHam]

@rduser for the image client case it appears the model does not have the inputs format specified in the model config.

Exception: unexpected input format FORMAT_NONE, expecting FORMAT_NCHW or FORMAT_NHWC

This needs to be specified to use the image client as is.

For the second error the number of elements is 783 but it should be 28281 = 784. This is why the HTTP server responds with the shape error.

Please make the necessary fix and try again.

[CoderHam]

I tried following config but get error saying mnist model failed

Can you share the exact error you are seeing?