[docs] Model docs #36469
[docs] Model docs #36469
Conversation
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
stevhliu
force-pushed
the
model-doc-design
branch
3 times, most recently
from
f353e9e to
8416e96
Compare
stevhliu
force-pushed
the
model-doc-design
branch
from
a172045 to
fcc9f21
Compare
stevhliu
force-pushed
the
model-doc-design
branch
2 times, most recently
from
9a3e0ca to
3d2a172
Compare
stevhliu
force-pushed
the
model-doc-design
branch
from
3d2a172 to
7c93f33
Compare
| Vision Transformer (ViT) attains excellent results compared to state-of-the-art convolutional networks while requiring | ||
| substantially fewer computational resources to train.* | ||
|
|
||
| <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/vit_architecture.jpg" |
There was a problem hiding this comment.
I noticed this PR removed the architecture image, could we please keep images in the docs?
They are stored at https://huggingface.co/datasets/huggingface/documentation-images.
cc @stevhliu
There was a problem hiding this comment.
Hi! While the image is a nice extra for users I'm not sure if it adds that much more value to the doc?
In general, I think it's better to move away from the older model card format which mirrored the model papers quite a bit (abstract + architecture image) to give it less of an academic/researchy vibe and something more immediately practical for developers. Users can click on the linked paper, which contains all that info, if they really want to learn more.
I think most users are more interested in just knowing how to use a model for inference and don't care too much about the details of the architecture. I also feel like some of the images are also a bit overwhelming for new users because there's a lot going on visually.
I'm not completely against the idea of removing the architecture image though and I'm open to adding it back 🙂
* initial * fix * fix * update * fix * fixes * quantization * attention mask visualizer * multimodal * small changes * fix code samples
🚧 WIP 🚧
Explores a new design for several of the most commonly visited model docs. It is pretty stripped down at the moment, but we can add back the essentials as needed (or not). The main idea is to only show developers what they need to know to use the model.