openvinotoolkit
[Good First Issue]: Improve logging for OpenVINO torch.compile feature
Context
Currently when developers are using torch.compile feature in openvino for their Pytorch models even with debug logging swtiched on they aren't able to see a lot of logs being printed for OpenVINO backend. So, the users are not able to understand if they model is running on OpenVINO or Native Pytorch.
What needs to be done?
Need to add verbose logging for OpenVINO Torch.Compile feature. The user needs to be able to tell if OpenVINO is being used and if so what are the subgraphs being delegated to openvino. Also, the user needs to be able to know when a model has fallen back to Native Pytorch instead of using OpenVINO.
Resources
Contact points
@suryasidd @ynimmaga @cavusmustafa
AI notice - Important!
We encourage contributors to use AI tools responsibly when creating Pull Requests. While AI can be a valuable aid, it is essential to ensure that your contributions meet the task requirements, build successfully, include relevant tests, and pass all linters. Submissions that do not meet these standards may be closed without warning to maintain the quality and integrity of the project. Please take the time to understand the changes you are proposing and their impact.
Ticket
No response
.take
I'm interested in this issue @suryasidd for GSoC 2026 preparation. I would like to work on this. Also could you point me to:
- The specific files/modules where torch.compile OpenVINO backend is implemented?
- The current logging infrastructure in OpenVINO?
- Any existing examples of verbose logging in other backends?
Hi I would like to work on this, could you please assign it to me.
@localvariable23 Are you working on this? I too have started to work on the issue, so please confirm your status.
.take
Hi! I have successfully set up OpenVINO + torch.compile on my machine and tested it. I would like to start working on improving logging for the OpenVINO torch.compile backend.
My initial plan: 1.Add debug/verbose-level logs in key execution points of the OpenVINO backend: a. When the OpenVINO backend is selected b. When specific subgraphs are delegated to OpenVINO c. When a fallback to native PyTorch happens 2. Use the existing logging utilities in openvino.torch and make logs visible when Python logging is set to DEBUG. 3. Add a small test to verify expected logs appear for a simple compiled model.
Please let me know if this direction aligns with your expectations or if you want the logs in specific locations.
