Non-zero status code returned while running Transpose node. Name:'Transpose_286' Status Message: CUDA error cudaErrorInvalidConfiguration:invalid configuration argument

[zll0000]

I trained a speech model with pytorch, and convert it to onnx model ,when I inference the result is right,
however I conver it to fp16 use the command
polygraphy convert test.onnx --fp-to-fp16 -o test_fp16.onnx
when i inference with short speech data, I can get right result,however When the speech data is long , I get the error as follows:
self._sess.run_with_iobinding(iobinding._iobinding, run_options)
RuntimeError: Error in execution: Non-zero status code returned while running Transpose node. Name:'Transpose_286' Status Message: CUDA error cudaErrorInvalidConfiguration:invalid configuration argument

@mtodd