-
Hi! Thank you very much for your excellent article.
I noticed that the backbone of image encoder you chose is vit-B16. Have you ever test others like vit-L14, vit-B32, etc? If you did, did vit-L work…
-
Hi there,
thank you for your amazing work! I have some questions regarding the loftr backbone. Do you use the pretrained loftr model for getting the feature and correspondences or train it from sc…
-
Could you please provide the backbones used for segmentation during training? I would want the one used for training VSSM small.
-
Good job! Have you tried to let go of the backbone of the SAM?
-
**Outdated checklist in the Backbone**
We recently received this message on helpdesk.
In Gbif, Begonia holostyla is corresponding to two different taxa.
The second one is "doubtful".
> The only acc…
-
Thanks for your work, that is amazing! Can you please share all backbone2d and backbone3d pretrained weights?
-
### Search before asking
- [X] I have searched the YOLOv8 [issues](https://github.com/ultralytics/ultralytics/issues) and [discussions](https://github.com/ultralytics/ultralytics/discussions) and fou…
-
Hi!
I found a problem with a name of insect vector while uploading this [dataset](https://www.gbif-uat.org/dataset/3b6a1c9e-6ac5-4043-b7df-97252d1e5d49).
The name is:
1. Culicoides festivipenni…
-
We want model parallelism to be easy to use across the library.
We should add a `get_layout_map()` implementation to all backbones. This should be mostly copy paste from the Gemma version, since al…
-
What is the meaning of this part when extracting features from the backbone? I cannot see any mention of this part in the paper.
```
conv_current = conv[:, 1:, :, :, :]
conv_prev = conv_current - c…