This model has been trained on 2M, from 3M to 5M. It was trained based on the best checkpoint from the latest training on 1M corrupted.
This model has been trained on 2M, from 3M to 5M. It was trained based on the best checkpoint from the latest training on 1M corrupted.