Type Error: Type 'tensor(int64)' of input parameter (relative_position) of operator (Min) in node (Min_2286) is invalid.

[vtfghs]

Describe the issue

When I tried to convert the model to onnx using torch, the above error occurred. How to fix it？

To reproduce

onnxruntime.capi.onnxruntime_pybind11_state.InvalidGraph: [ONNXRuntimeError] : 10 : INVALID_GRAPH : Load model failed:This is an invalid model.

Urgency

No response

Platform

Windows

OS Version

10

ONNX Runtime Installation

Released Package

ONNX Runtime Version or Commit ID

1.10.0

ONNX Runtime API

Python

Architecture

X64

Execution Provider

Default CPU

Execution Provider Library Version

No response

[wangyems]

int64 was supported since opset 12. Please consider using opset >= 12.
https://github.com/microsoft/onnxruntime/blob/main/docs/OperatorKernels.md#cpuexecutionprovider

Meanwhile the most recent ORT release is 1.14.1. Is it possible to upgrade the ORT version?