Hi! Thanks for sharing such an interesting paper and the corresponding codes. May I ask about the training resources and training time needed for both the pre-training and fine-tuning stages? Thanks in advance! Best regards