search

FudanNLPLAB / MouSi

Apache License 2.0
66 stars 0 forks source link

Which ConvNext did you use? #2

Closed Richar-Du closed 5 months ago

Richar-Du commented 5 months ago

I try to replicate the result of ConvNext, but the accuracy is much worse than the result in Table 2. Could you please clarify which pre-trained model did you use? I use the facebook/convnext-large-224.

Besides, could you please clarify the pre-trained models of other single-experts, like SAM, DINOv2, etc. Thanks in advance :)

cnxupupup commented 5 months ago

Thanks for the question. The use of the visual coder model is explained in the footnotes of the paper in arxiv.

Richar-Du commented 5 months ago

My apologies for missing it. Thank you.