How can I use the generated pte file to process my own data and predict the results? about executorch HOT 2 OPEN

PTE is the model represented in a way ExecuTorch runtime can consume it. The runtime can consume data / input tensors at runtime and run "forward" on the device.

Modify this to pass your inputs to the method.

https://github.com/pytorch/executorch/blob/main/examples/portable/executor_runner/executor_runner.cpp#L154-L171

from executorch.

I mean, before converting to pte file, my model input is processed by pytorch's dataloader

data_loader = DataLoader(dataset=dataset,
batch_size=batch_size,
shuffle=shuffle,
num_workers=0)

So in executor_runner.cpp, can I import libtorch to process data with pytorch's C++ API? Like below

#include<torch/torch.h>
#include
#include <executorch/extension/data_loader/file_data_loader.h>
#include <executorch/extension/evalue_util/print_evalue.h>

If this is possible, how can I use libtorch in executorch successfully?

from executorch.