Add metadata about GAN modules and update link to arxiv paper.

PiperOrigin-RevId: 204918179
rhaertel80 · Jul 18, 2018 · 17e813b · 17e813b
1 parent 102e763
commit 17e813b
Show file tree

Hide file tree

Showing 15 changed files with 90 additions and 30 deletions.
diff --git a/docs/modules/google/compare_gan/model_10_lsun_bedroom_resnet19/1.md b/docs/modules/google/compare_gan/model_10_lsun_bedroom_resnet19/1.md
@@ -1,6 +1,10 @@
 # Module google/compare_gan/model_10_lsun_bedroom_resnet19/1
 ResNet19 trained on LSUN Bedroom (FID: 40.36).
 
+<!-- module-type: image-generator -->
+<!-- network-architecture: ResNet19 -->
+<!-- dataset: LSUN Bedroom -->
+
 **Module URL:** [https://tfhub.dev/google/compare_gan/model_10_lsun_bedroom_resnet19/1](https://tfhub.dev/google/compare_gan/model_10_lsun_bedroom_resnet19/1)
 
 [![Open Colab notebok](../../../../images/open_in_colab.png)](https://colab.research.google.com/github/google/compare_gan/blob/master/compare_gan/src/tfhub_models.ipynb)
@@ -49,4 +53,4 @@ session.run([images, logits])
 ## References
 
 [1] Karol Kurach*, Mario Lucic*, Xiaohua Zhai, Marcin Michalski, Sylvain Gelly.
-[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://openreview.net/pdf?id=SkeFUsQNx7).
+[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://arxiv.org/abs/1807.04720).
diff --git a/docs/modules/google/compare_gan/model_11_cifar10_resnet_cifar/1.md b/docs/modules/google/compare_gan/model_11_cifar10_resnet_cifar/1.md
@@ -1,5 +1,9 @@
 # Module google/compare_gan/model_11_cifar10_resnet_cifar/1
-ResNet CIFAR trained on CIFAR10 (FID: 28.12).
+ResNet CIFAR trained on CIFAR-10 (FID: 28.12).
+
+<!-- module-type: image-generator -->
+<!-- network-architecture: ResNet CIFAR -->
+<!-- dataset: CIFAR-10 -->
 
 **Module URL:** [https://tfhub.dev/google/compare_gan/model_11_cifar10_resnet_cifar/1](https://tfhub.dev/google/compare_gan/model_11_cifar10_resnet_cifar/1)
 
@@ -16,7 +20,7 @@ View all available compare_gan modules in the [Colab notebook](https://colab.res
 
 #### Details
 
-* Dataset: CIFAR10
+* Dataset: CIFAR-10
 * Model: Non-saturating GAN
 * Architecture: ResNet CIFAR
 * Optimizer: Adam (lr=2.000e-04, beta1=0.500, beta2=0.999)
@@ -49,4 +53,4 @@ session.run([images, logits])
 ## References
 
 [1] Karol Kurach*, Mario Lucic*, Xiaohua Zhai, Marcin Michalski, Sylvain Gelly.
-[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://openreview.net/pdf?id=SkeFUsQNx7).
+[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://arxiv.org/abs/1807.04720).
diff --git a/docs/modules/google/compare_gan/model_12_cifar10_resnet_cifar/1.md b/docs/modules/google/compare_gan/model_12_cifar10_resnet_cifar/1.md
@@ -1,5 +1,9 @@
 # Module google/compare_gan/model_12_cifar10_resnet_cifar/1
-ResNet CIFAR trained on CIFAR10 (FID: 30.08).
+ResNet CIFAR trained on CIFAR-10 (FID: 30.08).
+
+<!-- module-type: image-generator -->
+<!-- network-architecture: ResNet CIFAR -->
+<!-- dataset: CIFAR-10 -->
 
 **Module URL:** [https://tfhub.dev/google/compare_gan/model_12_cifar10_resnet_cifar/1](https://tfhub.dev/google/compare_gan/model_12_cifar10_resnet_cifar/1)
 
@@ -16,7 +20,7 @@ View all available compare_gan modules in the [Colab notebook](https://colab.res
 
 #### Details
 
-* Dataset: CIFAR10
+* Dataset: CIFAR-10
 * Model: Non-saturating GAN
 * Architecture: ResNet CIFAR
 * Optimizer: Adam (lr=1.000e-04, beta1=0.500, beta2=0.999)
@@ -49,4 +53,4 @@ session.run([images, logits])
 ## References
 
 [1] Karol Kurach*, Mario Lucic*, Xiaohua Zhai, Marcin Michalski, Sylvain Gelly.
-[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://openreview.net/pdf?id=SkeFUsQNx7).
+[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://arxiv.org/abs/1807.04720).
diff --git a/docs/modules/google/compare_gan/model_13_cifar10_resnet_cifar/1.md b/docs/modules/google/compare_gan/model_13_cifar10_resnet_cifar/1.md
@@ -1,5 +1,9 @@
 # Module google/compare_gan/model_13_cifar10_resnet_cifar/1
-ResNet CIFAR trained on CIFAR10 (FID: 22.91).
+ResNet CIFAR trained on CIFAR-10 (FID: 22.91).
+
+<!-- module-type: image-generator -->
+<!-- network-architecture: ResNet CIFAR -->
+<!-- dataset: CIFAR-10 -->
 
 **Module URL:** [https://tfhub.dev/google/compare_gan/model_13_cifar10_resnet_cifar/1](https://tfhub.dev/google/compare_gan/model_13_cifar10_resnet_cifar/1)
 
@@ -16,7 +20,7 @@ View all available compare_gan modules in the [Colab notebook](https://colab.res
 
 #### Details
 
-* Dataset: CIFAR10
+* Dataset: CIFAR-10
 * Model: Non-saturating GAN
 * Architecture: ResNet CIFAR
 * Optimizer: Adam (lr=2.000e-04, beta1=0.500, beta2=0.999)
@@ -49,4 +53,4 @@ session.run([images, logits])
 ## References
 
 [1] Karol Kurach*, Mario Lucic*, Xiaohua Zhai, Marcin Michalski, Sylvain Gelly.
-[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://openreview.net/pdf?id=SkeFUsQNx7).
+[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://arxiv.org/abs/1807.04720).
diff --git a/docs/modules/google/compare_gan/model_14_cifar10_resnet_cifar/1.md b/docs/modules/google/compare_gan/model_14_cifar10_resnet_cifar/1.md
@@ -1,5 +1,9 @@
 # Module google/compare_gan/model_14_cifar10_resnet_cifar/1
-ResNet CIFAR trained on CIFAR10 (FID: 23.22).
+ResNet CIFAR trained on CIFAR-10 (FID: 23.22).
+
+<!-- module-type: image-generator -->
+<!-- network-architecture: ResNet CIFAR -->
+<!-- dataset: CIFAR-10 -->
 
 **Module URL:** [https://tfhub.dev/google/compare_gan/model_14_cifar10_resnet_cifar/1](https://tfhub.dev/google/compare_gan/model_14_cifar10_resnet_cifar/1)
 
@@ -16,7 +20,7 @@ View all available compare_gan modules in the [Colab notebook](https://colab.res
 
 #### Details
 
-* Dataset: CIFAR10
+* Dataset: CIFAR-10
 * Model: Non-saturating GAN
 * Architecture: ResNet CIFAR
 * Optimizer: Adam (lr=2.000e-04, beta1=0.500, beta2=0.999)
@@ -49,4 +53,4 @@ session.run([images, logits])
 ## References
 
 [1] Karol Kurach*, Mario Lucic*, Xiaohua Zhai, Marcin Michalski, Sylvain Gelly.
-[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://openreview.net/pdf?id=SkeFUsQNx7).
+[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://arxiv.org/abs/1807.04720).
diff --git a/docs/modules/google/compare_gan/model_15_cifar10_resnet_cifar/1.md b/docs/modules/google/compare_gan/model_15_cifar10_resnet_cifar/1.md
@@ -1,5 +1,9 @@
 # Module google/compare_gan/model_15_cifar10_resnet_cifar/1
-ResNet CIFAR trained on CIFAR10 (FID: 22.73).
+ResNet CIFAR trained on CIFAR-10 (FID: 22.73).
+
+<!-- module-type: image-generator -->
+<!-- network-architecture: ResNet CIFAR -->
+<!-- dataset: CIFAR-10 -->
 
 **Module URL:** [https://tfhub.dev/google/compare_gan/model_15_cifar10_resnet_cifar/1](https://tfhub.dev/google/compare_gan/model_15_cifar10_resnet_cifar/1)
 
@@ -16,7 +20,7 @@ View all available compare_gan modules in the [Colab notebook](https://colab.res
 
 #### Details
 
-* Dataset: CIFAR10
+* Dataset: CIFAR-10
 * Model: Non-saturating GAN
 * Architecture: ResNet CIFAR
 * Optimizer: Adam (lr=2.000e-04, beta1=0.500, beta2=0.999)
@@ -49,4 +53,4 @@ session.run([images, logits])
 ## References
 
 [1] Karol Kurach*, Mario Lucic*, Xiaohua Zhai, Marcin Michalski, Sylvain Gelly.
-[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://openreview.net/pdf?id=SkeFUsQNx7).
+[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://arxiv.org/abs/1807.04720).
diff --git a/docs/modules/google/compare_gan/model_1_celebahq128_resnet19/1.md b/docs/modules/google/compare_gan/model_1_celebahq128_resnet19/1.md
@@ -1,6 +1,10 @@
 # Module google/compare_gan/model_1_celebahq128_resnet19/1
 ResNet19 trained on CelebA HQ (128x128) (FID: 34.29).
 
+<!-- module-type: image-generator -->
+<!-- network-architecture: ResNet19 -->
+<!-- dataset: CelebA HQ -->
+
 **Module URL:** [https://tfhub.dev/google/compare_gan/model_1_celebahq128_resnet19/1](https://tfhub.dev/google/compare_gan/model_1_celebahq128_resnet19/1)
 
 [![Open Colab notebok](../../../../images/open_in_colab.png)](https://colab.research.google.com/github/google/compare_gan/blob/master/compare_gan/src/tfhub_models.ipynb)
@@ -16,7 +20,7 @@ View all available compare_gan modules in the [Colab notebook](https://colab.res
 
 #### Details
 
-* Dataset: CelebA HQ (128x128)
+* Dataset: CelebA HQ
 * Model: Non-saturating GAN
 * Architecture: ResNet19
 * Optimizer: Adam (lr=3.381e-05, beta1=0.375, beta2=0.998)
@@ -49,4 +53,4 @@ session.run([images, logits])
 ## References
 
 [1] Karol Kurach*, Mario Lucic*, Xiaohua Zhai, Marcin Michalski, Sylvain Gelly.
-[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://openreview.net/pdf?id=SkeFUsQNx7).
+[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://arxiv.org/abs/1807.04720).
diff --git a/docs/modules/google/compare_gan/model_2_celebahq128_resnet19/1.md b/docs/modules/google/compare_gan/model_2_celebahq128_resnet19/1.md
@@ -1,6 +1,10 @@
 # Module google/compare_gan/model_2_celebahq128_resnet19/1
 ResNet19 trained on CelebA HQ (128x128) (FID: 35.85).
 
+<!-- module-type: image-generator -->
+<!-- network-architecture: ResNet19 -->
+<!-- dataset: CelebA HQ -->
+
 **Module URL:** [https://tfhub.dev/google/compare_gan/model_2_celebahq128_resnet19/1](https://tfhub.dev/google/compare_gan/model_2_celebahq128_resnet19/1)
 
 [![Open Colab notebok](../../../../images/open_in_colab.png)](https://colab.research.google.com/github/google/compare_gan/blob/master/compare_gan/src/tfhub_models.ipynb)
@@ -16,7 +20,7 @@ View all available compare_gan modules in the [Colab notebook](https://colab.res
 
 #### Details
 
-* Dataset: CelebA HQ (128x128)
+* Dataset: CelebA HQ
 * Model: Non-saturating GAN
 * Architecture: ResNet19
 * Optimizer: Adam (lr=1.000e-04, beta1=0.500, beta2=0.999)
@@ -49,4 +53,4 @@ session.run([images, logits])
 ## References
 
 [1] Karol Kurach*, Mario Lucic*, Xiaohua Zhai, Marcin Michalski, Sylvain Gelly.
-[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://openreview.net/pdf?id=SkeFUsQNx7).
+[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://arxiv.org/abs/1807.04720).
diff --git a/docs/modules/google/compare_gan/model_3_lsun_bedroom_resnet19/1.md b/docs/modules/google/compare_gan/model_3_lsun_bedroom_resnet19/1.md
@@ -1,6 +1,10 @@
 # Module google/compare_gan/model_3_lsun_bedroom_resnet19/1
 ResNet19 trained on LSUN Bedroom (FID: 102.74).
 
+<!-- module-type: image-generator -->
+<!-- network-architecture: ResNet19 -->
+<!-- dataset: LSUN Bedroom -->
+
 **Module URL:** [https://tfhub.dev/google/compare_gan/model_3_lsun_bedroom_resnet19/1](https://tfhub.dev/google/compare_gan/model_3_lsun_bedroom_resnet19/1)
 
 [![Open Colab notebok](../../../../images/open_in_colab.png)](https://colab.research.google.com/github/google/compare_gan/blob/master/compare_gan/src/tfhub_models.ipynb)
@@ -49,4 +53,4 @@ session.run([images, logits])
 ## References
 
 [1] Karol Kurach*, Mario Lucic*, Xiaohua Zhai, Marcin Michalski, Sylvain Gelly.
-[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://openreview.net/pdf?id=SkeFUsQNx7).
+[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://arxiv.org/abs/1807.04720).
diff --git a/docs/modules/google/compare_gan/model_4_lsun_bedroom_resnet19/1.md b/docs/modules/google/compare_gan/model_4_lsun_bedroom_resnet19/1.md
@@ -1,6 +1,10 @@
 # Module google/compare_gan/model_4_lsun_bedroom_resnet19/1
 ResNet19 trained on LSUN Bedroom (FID: 112.92).
 
+<!-- module-type: image-generator -->
+<!-- network-architecture: ResNet19 -->
+<!-- dataset: LSUN Bedroom -->
+
 **Module URL:** [https://tfhub.dev/google/compare_gan/model_4_lsun_bedroom_resnet19/1](https://tfhub.dev/google/compare_gan/model_4_lsun_bedroom_resnet19/1)
 
 [![Open Colab notebok](../../../../images/open_in_colab.png)](https://colab.research.google.com/github/google/compare_gan/blob/master/compare_gan/src/tfhub_models.ipynb)
@@ -49,4 +53,4 @@ session.run([images, logits])
 ## References
 
 [1] Karol Kurach*, Mario Lucic*, Xiaohua Zhai, Marcin Michalski, Sylvain Gelly.
-[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://openreview.net/pdf?id=SkeFUsQNx7).
+[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://arxiv.org/abs/1807.04720).
diff --git a/docs/modules/google/compare_gan/model_5_celebahq128_resnet19/1.md b/docs/modules/google/compare_gan/model_5_celebahq128_resnet19/1.md
@@ -1,6 +1,10 @@
 # Module google/compare_gan/model_5_celebahq128_resnet19/1
 ResNet19 trained on CelebA HQ (128x128) (FID: 30.02).
 
+<!-- module-type: image-generator -->
+<!-- network-architecture: ResNet19 -->
+<!-- dataset: CelebA HQ -->
+
 **Module URL:** [https://tfhub.dev/google/compare_gan/model_5_celebahq128_resnet19/1](https://tfhub.dev/google/compare_gan/model_5_celebahq128_resnet19/1)
 
 [![Open Colab notebok](../../../../images/open_in_colab.png)](https://colab.research.google.com/github/google/compare_gan/blob/master/compare_gan/src/tfhub_models.ipynb)
@@ -16,7 +20,7 @@ View all available compare_gan modules in the [Colab notebook](https://colab.res
 
 #### Details
 
-* Dataset: CelebA HQ (128x128)
+* Dataset: CelebA HQ
 * Model: Non-saturating GAN
 * Architecture: ResNet19
 * Optimizer: Adam (lr=1.000e-04, beta1=0.500, beta2=0.999)
@@ -49,4 +53,4 @@ session.run([images, logits])
 ## References
 
 [1] Karol Kurach*, Mario Lucic*, Xiaohua Zhai, Marcin Michalski, Sylvain Gelly.
-[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://openreview.net/pdf?id=SkeFUsQNx7).
+[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://arxiv.org/abs/1807.04720).
diff --git a/docs/modules/google/compare_gan/model_6_celebahq128_resnet19/1.md b/docs/modules/google/compare_gan/model_6_celebahq128_resnet19/1.md
@@ -1,6 +1,10 @@
 # Module google/compare_gan/model_6_celebahq128_resnet19/1
 ResNet19 trained on CelebA HQ (128x128) (FID: 32.05).
 
+<!-- module-type: image-generator -->
+<!-- network-architecture: ResNet19 -->
+<!-- dataset: CelebA HQ -->
+
 **Module URL:** [https://tfhub.dev/google/compare_gan/model_6_celebahq128_resnet19/1](https://tfhub.dev/google/compare_gan/model_6_celebahq128_resnet19/1)
 
 [![Open Colab notebok](../../../../images/open_in_colab.png)](https://colab.research.google.com/github/google/compare_gan/blob/master/compare_gan/src/tfhub_models.ipynb)
@@ -16,7 +20,7 @@ View all available compare_gan modules in the [Colab notebook](https://colab.res
 
 #### Details
 
-* Dataset: CelebA HQ (128x128)
+* Dataset: CelebA HQ
 * Model: Non-saturating GAN
 * Architecture: ResNet19
 * Optimizer: Adam (lr=1.000e-04, beta1=0.500, beta2=0.999)
@@ -49,4 +53,4 @@ session.run([images, logits])
 ## References
 
 [1] Karol Kurach*, Mario Lucic*, Xiaohua Zhai, Marcin Michalski, Sylvain Gelly.
-[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://openreview.net/pdf?id=SkeFUsQNx7).
+[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://arxiv.org/abs/1807.04720).
diff --git a/docs/modules/google/compare_gan/model_7_lsun_bedroom_resnet19/1.md b/docs/modules/google/compare_gan/model_7_lsun_bedroom_resnet19/1.md
@@ -1,6 +1,10 @@
 # Module google/compare_gan/model_7_lsun_bedroom_resnet19/1
 ResNet19 trained on LSUN Bedroom (FID: 41.60).
 
+<!-- module-type: image-generator -->
+<!-- network-architecture: ResNet19 -->
+<!-- dataset: LSUN Bedroom -->
+
 **Module URL:** [https://tfhub.dev/google/compare_gan/model_7_lsun_bedroom_resnet19/1](https://tfhub.dev/google/compare_gan/model_7_lsun_bedroom_resnet19/1)
 
 [![Open Colab notebok](../../../../images/open_in_colab.png)](https://colab.research.google.com/github/google/compare_gan/blob/master/compare_gan/src/tfhub_models.ipynb)
@@ -49,4 +53,4 @@ session.run([images, logits])
 ## References
 
 [1] Karol Kurach*, Mario Lucic*, Xiaohua Zhai, Marcin Michalski, Sylvain Gelly.
-[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://openreview.net/pdf?id=SkeFUsQNx7).
+[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://arxiv.org/abs/1807.04720).
diff --git a/docs/modules/google/compare_gan/model_8_lsun_bedroom_resnet19/1.md b/docs/modules/google/compare_gan/model_8_lsun_bedroom_resnet19/1.md
@@ -1,6 +1,10 @@
 # Module google/compare_gan/model_8_lsun_bedroom_resnet19/1
 ResNet19 trained on LSUN Bedroom (FID: 42.51).
 
+<!-- module-type: image-generator -->
+<!-- network-architecture: ResNet19 -->
+<!-- dataset: LSUN Bedroom -->
+
 **Module URL:** [https://tfhub.dev/google/compare_gan/model_8_lsun_bedroom_resnet19/1](https://tfhub.dev/google/compare_gan/model_8_lsun_bedroom_resnet19/1)
 
 [![Open Colab notebok](../../../../images/open_in_colab.png)](https://colab.research.google.com/github/google/compare_gan/blob/master/compare_gan/src/tfhub_models.ipynb)
@@ -49,4 +53,4 @@ session.run([images, logits])
 ## References
 
 [1] Karol Kurach*, Mario Lucic*, Xiaohua Zhai, Marcin Michalski, Sylvain Gelly.
-[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://openreview.net/pdf?id=SkeFUsQNx7).
+[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://arxiv.org/abs/1807.04720).
diff --git a/docs/modules/google/compare_gan/model_9_celebahq128_resnet19/1.md b/docs/modules/google/compare_gan/model_9_celebahq128_resnet19/1.md
@@ -1,6 +1,10 @@
 # Module google/compare_gan/model_9_celebahq128_resnet19/1
 ResNet19 trained on CelebA HQ (128x128) (FID: 29.13).
 
+<!-- module-type: image-generator -->
+<!-- network-architecture: ResNet19 -->
+<!-- dataset: CelebA HQ -->
+
 **Module URL:** [https://tfhub.dev/google/compare_gan/model_9_celebahq128_resnet19/1](https://tfhub.dev/google/compare_gan/model_9_celebahq128_resnet19/1)
 
 [![Open Colab notebok](../../../../images/open_in_colab.png)](https://colab.research.google.com/github/google/compare_gan/blob/master/compare_gan/src/tfhub_models.ipynb)
@@ -16,7 +20,7 @@ View all available compare_gan modules in the [Colab notebook](https://colab.res
 
 #### Details
 
-* Dataset: CelebA HQ (128x128)
+* Dataset: CelebA HQ
 * Model: Non-saturating GAN
 * Architecture: ResNet19
 * Optimizer: Adam (lr=1.000e-04, beta1=0.500, beta2=0.900)
@@ -49,4 +53,4 @@ session.run([images, logits])
 ## References
 
 [1] Karol Kurach*, Mario Lucic*, Xiaohua Zhai, Marcin Michalski, Sylvain Gelly.
-[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://openreview.net/pdf?id=SkeFUsQNx7).
+[The GAN Landscape: Losses, Architectures, Regularization, and Normalization](https://arxiv.org/abs/1807.04720).