The performance cannot be reproduced

[645873229]

The performance cannot be reproduced. The experiment was conducted on Shanghaitech A, and the mae of the method using the token is 70.4, and the method using the gap is only 74. All experiments performed in accordance with the experimental hyperparameters

[dk-liang]

Do you use the official code and load the pre-trained model from Imagenet?

[645873229]

Do you use the official code and load the pre-trained model from Imagenet?

Yes, the data augmentation is missing in the official code and I added it, but the performance of the transcrowd_gap is still not reproduced