How were the two models being created?

[dogma346]

Hi, Sidak. My name is Terai, and I am an undergraduate student studying informatics engineering.

I have read your paper (titled as Model Fusion via Optimal Transport) published on NIPS'20. After reading through the paper, I have tried to reproduce the simulation results using the codes publicly available on GitHub, but I have two simple questions: How were the two models (used in Figure 2 of the paper) being created? How can I leverage your codes to retrain the models by myself?

I know you are busy, but I would greatly appreciate it if you could help me. Thanks.

[XueY-L]

Hi, Sidak. My name is Terai, and I am an undergraduate student studying informatics engineering.

I have read your paper (titled as Model Fusion via Optimal Transport) published on NIPS'20. After reading through the paper, I have tried to reproduce the simulation results using the codes publicly available on GitHub, but I have two simple questions: How were the two models (used in Figure 2 of the paper) being created? How can I leverage your codes to retrain the models by myself?

I know you are busy, but I would greatly appreciate it if you could help me. Thanks.

Hi, Terai. I have also tried to reproduce the results. The resnet model may be trained using cifar/models/resnet.py. Please note that this model structure is different from that in torchvision.models. And the training hyperparameters are listed in cifar/hyperparameters. You can use train_cifar_models.py to train source models.

I have trained resnet models with BN layer and linear layer bias, while the provided models only contain the conv weights and linear weights. So I am not sure the actual model architecture during training.