Update README.md
Browse files
README.md
CHANGED
|
@@ -15,7 +15,7 @@ Pre-trained checkpoint for **"LookWhen? Fast Video Recognition by Learning When,
|
|
| 15 |
and What to Compute"** ([arXiv:2605.06809](https://arxiv.org/abs/2605.06809)).
|
| 16 |
|
| 17 |
LookWhen splits video recognition in two: a shallow **selector** scores every patch of a
|
| 18 |
-
2x-downscaled clip, and a deep **extractor** processes only the top-K of them while still
|
| 19 |
predicting features of the whole video.
|
| 20 |
|
| 21 |
This is the model used throughout the paper: ViT-B/16, pre-trained for 20 epochs on
|
|
|
|
| 15 |
and What to Compute"** ([arXiv:2605.06809](https://arxiv.org/abs/2605.06809)).
|
| 16 |
|
| 17 |
LookWhen splits video recognition in two: a shallow **selector** scores every patch of a
|
| 18 |
+
2x-downscaled (2x in each dimension) clip, and a deep **extractor** processes only the top-K of them while still
|
| 19 |
predicting features of the whole video.
|
| 20 |
|
| 21 |
This is the model used throughout the paper: ViT-B/16, pre-trained for 20 epochs on
|