Open
Description
Describe the issue
Hi dear all,
I was wondering which version of torch with cuda 12.4 is compatible in order to set OPSET=23 to export my model, since it keeps me telling me'RuntimeError: Unsupported onnx_opset_version: 23'. I am using 12.4 CUDA, so I was wondering if you could me lead me to a solution since my code need dynamic attetion part to be exported, otherwise when loading model from 18 version keeps me telling reshape is wrong with different input. Thanks a lot
To reproduce
To reproduce just take any attetion network, given that which version of torch is compatible with onnx in order to be able to export the whole model correctly and not getting any OPSET=23 not existent?
Urgency
No response
Platform
Windows
OS Version
Linux
ONNX Runtime Installation
Built from Source
ONNX Runtime Version or Commit ID
1.18
ONNX Runtime API
Python
Architecture
X64
Execution Provider
CUDA
Execution Provider Library Version
CUDA 12.4