Thank you for your amazing work. I noticed that you trained all models for about 2, 000, 000 iterations. It may takes too much time to train the model from scratch. So can you release some checkpoints or provide some training details such as hardware and training time?
Thank you for your amazing work. I noticed that you trained all models for about 2, 000, 000 iterations. It may takes too much time to train the model from scratch. So can you release some checkpoints or provide some training details such as hardware and training time?