Yi-Fan Zhang

I use 4 GPU for training coco to 90k iters, and it drops to 34.7AP without modifying any configure

Because our download file is in the main folder not in the domainbed.scripts folder, I suggest you use download.py directly. Just set your own dir. I will also fix fix...

> python download.py --data_dir=/my/datasets/path return No such file or directory: '/my/datasets/path\terra_incognita\terra_incognita_images.tar.gz' Please make sure the path /my/datasets/path exists and is correct. It seems unfamiliar that the first part uses '/'...

we indeed included such a comparison in our study. Specifically, you can find the relevant findings in Table 5 of our paper, along with a detailed discussion in Section 4.3....

Regarding your concern about the dataset size and memory management, we acknowledge that larger datasets can pose significant challenges in terms of memory utilization. While we have successfully tested our...

For the average method, you can simply aggregate the results from different experts and compute the average. Regarding the MOE (Mixture of Experts), the implementation is akin to gating. The...

It is unreasonable, the ECL dataset also contains more than 100,000 time steps and a large number of channels, works well with the model.

where is the code from？

Thanks for pointing that out! I'm not entirely clear on the issue you're describing, though. Can you explain the problem a bit more? I'd really appreciate it if you could...