About the training of DPC on 230K Surreal datasets

[2019EPWL]

Hi, I see that you show the DPC results on the 230k surreal dataset in Table 3, but the DPC repo only gives the checkpoint trained at 2k, so did you retrain the DPC yourself? But I retrain DPC on the 230k surreal dataset, and the effect is not as good as in the paper, so can you provide the checkpoint of your training on 230k?