Skip to content

Commit

Permalink
add evonorm links
Browse files Browse the repository at this point in the history
  • Loading branch information
leondgarse committed Apr 5, 2022
1 parent 53b322c commit 6975b9e
Show file tree
Hide file tree
Showing 3 changed files with 4 additions and 2 deletions.
3 changes: 2 additions & 1 deletion keras_cv_attention_models/attention_layers/__init__.py
Original file line number Diff line number Diff line change
Expand Up @@ -120,7 +120,8 @@
"""

EvoNormalization.__doc__ = """
Keras implementation of [PDF 2004.02967 Evolving Normalization-Activation Layers](https://arxiv.org/pdf/2004.02967.pdf).
Keras implementation of [evonorm](https://github.com/tensorflow/tpu/blob/master/models/official/resnet/resnet_model.py).
Paper [PDF 2004.02967 Evolving Normalization-Activation Layers](https://arxiv.org/pdf/2004.02967.pdf).
Args:
nonlinearity: boolean value if apply non-linearity calculation, default True.
Expand Down
1 change: 1 addition & 0 deletions keras_cv_attention_models/common_layers.py
Original file line number Diff line number Diff line change
Expand Up @@ -63,6 +63,7 @@ def activation_by_name(inputs, activation="relu", name=None):
@tf.keras.utils.register_keras_serializable(package="kecamCommon")
class EvoNormalization(tf.keras.layers.Layer):
def __init__(self, nonlinearity=True, num_groups=-1, zero_gamma=False, momentum=0.99, epsilon=0.001, data_format="auto", **kwargs):
# [evonorm](https://github.com/tensorflow/tpu/blob/master/models/official/resnet/resnet_model.py)
# EVONORM_B0: nonlinearity=True, num_groups=-1
# EVONORM_S0: nonlinearity=True, num_groups > 0
# EVONORM_B0 / EVONORM_S0 linearity: nonlinearity=False, num_groups=-1
Expand Down
2 changes: 1 addition & 1 deletion keras_cv_attention_models/resnet_family/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -116,7 +116,7 @@
| RegNetY320 | 145.05M | 224 | 82.5 | [regnety_320_imagenet.h5](https://github.com/leondgarse/keras_cv_attention_models/releases/download/resnet_family/regnety_320_imagenet.h5) |
## RegNetZ
- Defined and model weights loaded from [Github timm/models/byobnet.py](https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/byobnet.py).
- Related paper [PDF 2004.02967 Evolving Normalization-Activation Layers](https://arxiv.org/pdf/2004.02967.pdf).
- [evonorm](https://github.com/tensorflow/tpu/blob/master/models/official/resnet/resnet_model.py), paper [PDF 2004.02967 Evolving Normalization-Activation Layers](https://arxiv.org/pdf/2004.02967.pdf).

| Model | Params | Image resolution | Top1 Acc | Download |
| -------------- | ------ | ---------------- | -------- | -------- |
Expand Down

0 comments on commit 6975b9e

Please sign in to comment.