Conversion fails randomly on models with multiple inputs about nobuco

[AlexanderLutsenko]

Okay, I ran the script on a Windows machine, can confirm it's a bug. Thanks for reporting!

[AlexanderLutsenko]

Hi! I just launched the script 50 times over, no issues found. Could that be a broken environment? A new Python (3.11)? Would you care to try it on a fresh env. with the following dependencies:

python 3.9
pip install torch torchvision torchaudio tensorflow==2.15 nobuco

[mmjamm]

Thank you for the fast response time!

I recreated the env on another machine using both your suggested dependencies and also what I had used originally and was not able to reproduce the issue either way.

I can still reproduce the problem on the original offending machine after recreating envs there, but it feels likely that it's a local problem. (The working configuration is on Linux and the problematic one is on Windows so it might also be a Windows problem, but my gut feeling is that it's some local problem in my environment.)

If there's something you'd like me to try, I can.
I'm also ok to close this issue, since I wasn't able to reproduce it on Linux.

[AlexanderLutsenko]

Does a fresh environment work on Windows, though?

[mmjamm]

I couldn't get a fresh environment to work on Windows.

[AlexanderLutsenko]

I think I fixed the bug in v0.14.0. @mmjamm Let me know if it works for you!

[mmjamm]

I think I fixed the bug in v0.14.0. @mmjamm Let me know if it works for you!

It works for me. Tested with v0.14.4.
Thank you very much!