amyeroberts HF staff commited on
Commit
dd659a0
1 Parent(s): 84e6627

Update dropout value

Browse files

Dropout prob in the transformers model layers is defined as the probability of an activation being set to 0, in the original MobileNet repo it's defined as the keep probability.

See:
- https://github.com/huggingface/transformers/issues/22767
- https://github.com/tensorflow/models/blob/ad32e81e31232675319d5572b78bc196216fd52e/research/slim/nets/mobilenet_v1.py#L306

Files changed (1) hide show
  1. config.json +1 -1
config.json CHANGED
@@ -2,7 +2,7 @@
2
  "architectures": [
3
  "MobileNetV2ForImageClassification"
4
  ],
5
- "classifier_dropout_prob": 0.8,
6
  "depth_divisible_by": 8,
7
  "depth_multiplier": 1.0,
8
  "expand_ratio": 6,
 
2
  "architectures": [
3
  "MobileNetV2ForImageClassification"
4
  ],
5
+ "classifier_dropout_prob": 0.2,
6
  "depth_divisible_by": 8,
7
  "depth_multiplier": 1.0,
8
  "expand_ratio": 6,