Closed hwidong-na closed 3 years ago
This is the configuration used, but it would take over 10 days to train on a single V100 GPU. We trained this using distributed training which is currently undergoing some testing before the public release (#61).
See #61
Hi,
Thank you for your effort to provide improved baseline model. It would be helpful for me to know more information e.g. hyperparameters for training the model.
Best, Hwidong