Mirrors_Framework/deepmind-research
1
0
Fork 0
mirror of https://github.com/google-deepmind/deepmind-research.git synced 2026-05-28 02:35:47 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

Fix Perceiver IO author list, add correct ImageNet paper section.

Browse Source 
PiperOrigin-RevId: 388423720
This commit is contained in:
Drew Jaegle
2021-08-03 12:24:01 +01:00
committed by alimuldal
parent 887a4fcda6
commit 21084c8489
1 changed files with 4 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
perceiver/README.md
+4 -4
View File
@@ -70,7 +70,7 @@ We also provide an example training script to train a Perceiver IO model for
ImageNet classification. ImageNet classification.
The provided hyperparameters are the settings used to train Perceiver IO The provided hyperparameters are the settings used to train Perceiver IO
with 2D Fourier position encodings, as described in with 2D Fourier position encodings, as described in
section 4.5 and supplemental section I.1 of the paper [2]. section 4.5 and supplemental section H of the paper [2].
To run the script locally and train a miniature Perceiver model, To run the script locally and train a miniature Perceiver model,
run: `perceiver/train/launch_local.sh`. run: `perceiver/train/launch_local.sh`.
@@ -107,9 +107,9 @@ João Carreira.
https://arxiv.org/abs/2103.03206 https://arxiv.org/abs/2103.03206
[2] Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch, [2] Andrew Jaegle, Sebastian Borgeaud, Jean-Baptiste Alayrac, Carl Doersch,
Catalin Ionescu, David Ding, Skanda Koppula, Andrew Brock, Evan Shelhamer, Catalin Ionescu, David Ding, Skanda Koppula, Daniel Zoran, Andrew Brock,
Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman, Oriol Vinyals, Evan Shelhamer, Olivier Hénaff, Matthew M. Botvinick, Andrew Zisserman,
João Carreira. Oriol Vinyals, João Carreira.
*Perceiver IO: A General Architecture for Structured Inputs & Outputs*. *Perceiver IO: A General Architecture for Structured Inputs & Outputs*.
arXiv, 2021. arXiv, 2021.
https://arxiv.org/abs/2107.14795 https://arxiv.org/abs/2107.14795