Error in loading pre-trained models for minigames

[XinyiYS]

Hi just wanted to reproduce the results reported by downloading the zip files from releases

I ran into two issues:

1. In the unzipped folders, there seems to be missing the config.gin files
2. In an attempt to resolve the first issue, I started the training for the corresponding minigame and interrupted to get the config.gin file. But loading the model checkpoints seems to give another problem, giving the error message: Restoring from checkpoint failed. This is most likely due to a Variable name or other graph key that is missing from the checkpoint. Could it be due to the version of tensorflow?

Thanks again!

[inoryy]

Hello,

1. A dedicated config shouldn't be necessary as the model weights were trained with the provided configs (although that might have changed, will need to double check).
2. This is indeed probably due to updates to TF version and I will have to re-train the models from scratch. In the meantime you can use the same version of code and deps as the one used when weights were released.

[XinyiYS]

Got it. Was just wondering if there would be a hack or workaround of the error. Thanks for the response.