python3+pytorch1.6 reasoning failed

[cccccv-cm]

Hello, excuse me. For some special reason I have to use python3.6+pytorch1.6 for inference on mapnet, but to my surprise the results in this case are worse than inference with the environment in the environment.yml file Many, what is the reason for this, if you can answer, I will be very grateful.

[samarth-robo]

Hi @cccccv-cm you are right that in the past we have observed that changing the PyTorch version affects reproducibility: https://github.com/NVlabs/geomapnet/issues/36. I don't have the ability to support newer and newer versions. So one option would be for you to re-train the model with the new version of PyTorch. Training code is included in this repository, and many others have reproduced the training results.

[cccccv-cm]

Hi @cccccv-cm you are right that in the past we have observed that changing the PyTorch version affects reproducibility: #36. I don't have the ability to support newer and newer versions. So one option would be for you to re-train the model with the new version of PyTorch. Training code is included in this repository, and many others have reproduced the training results. Oh I see, I'll try to retrain with a different version. Thanks for your advice and thanks for your work