PrakhAI commited on
Commit
e039ba0
·
1 Parent(s): d6f77c9

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +7 -3
README.md CHANGED
@@ -13,7 +13,7 @@ Try out this model [here](https://huggingface.co/spaces/PrakhAI/AIPlane).
13
  | ![generated_1691259071.png](https://cdn-uploads.huggingface.co/production/uploads/649f9483d76ca0fe679011c2/DNio2mes1414p6cgm7K62.png) | ![image.png](https://cdn-uploads.huggingface.co/production/uploads/649f9483d76ca0fe679011c2/4Sp33Hl9JK2cfHzBXHXfh.png) |
14
 
15
  # Training Progression
16
- <video width="50%" controls>
17
  <source src="https://cdn-uploads.huggingface.co/production/uploads/649f9483d76ca0fe679011c2/qFlnTITZwS3DSTxLp0Oa8.mp4" type="video/mp4">
18
  </video>
19
 
@@ -32,7 +32,7 @@ Spectral Normalization is a technique suggested for training GANs in [this paper
32
 
33
  It aims to make the Critic's (Discriminator's) outputs mathematically continuous w.r.t. the space of input images, avoiding exploding gradients.
34
 
35
- It works very well in practice to stabilize the training of the GAN, as demonstrated by the example below (comparison at equivalent points during training):
36
 
37
  | Batch Normalization | Spectral Normalization |
38
  | ----------- | ------------ |
@@ -46,4 +46,8 @@ For 32x32 images of Airplanes, even a short initial round of Progressive Growing
46
 
47
  | Flat Growing | Progressive Growing |
48
  | ----------- | ------------ |
49
- | ![image.png](https://cdn-uploads.huggingface.co/production/uploads/649f9483d76ca0fe679011c2/QnTET-5ae_0x11CcXeWgR.png) | ![image.png](https://cdn-uploads.huggingface.co/production/uploads/649f9483d76ca0fe679011c2/F8q4y2vshssfdc70jH_X2.png) |
 
 
 
 
 
13
  | ![generated_1691259071.png](https://cdn-uploads.huggingface.co/production/uploads/649f9483d76ca0fe679011c2/DNio2mes1414p6cgm7K62.png) | ![image.png](https://cdn-uploads.huggingface.co/production/uploads/649f9483d76ca0fe679011c2/4Sp33Hl9JK2cfHzBXHXfh.png) |
14
 
15
  # Training Progression
16
+ <video width="25%" controls>
17
  <source src="https://cdn-uploads.huggingface.co/production/uploads/649f9483d76ca0fe679011c2/qFlnTITZwS3DSTxLp0Oa8.mp4" type="video/mp4">
18
  </video>
19
 
 
32
 
33
  It aims to make the Critic's (Discriminator's) outputs mathematically continuous w.r.t. the space of input images, avoiding exploding gradients.
34
 
35
+ Spectral Normalization works very well in practice to stabilize the training of the GAN, as demonstrated by the example below (comparison at equivalent points during training):
36
 
37
  | Batch Normalization | Spectral Normalization |
38
  | ----------- | ------------ |
 
46
 
47
  | Flat Growing | Progressive Growing |
48
  | ----------- | ------------ |
49
+ | ![image.png](https://cdn-uploads.huggingface.co/production/uploads/649f9483d76ca0fe679011c2/QnTET-5ae_0x11CcXeWgR.png) | ![image.png](https://cdn-uploads.huggingface.co/production/uploads/649f9483d76ca0fe679011c2/F8q4y2vshssfdc70jH_X2.png) |
50
+
51
+ The generator for this model generates 4x4, 8x8, 16x16 and 32x32 images, which form the inputs for the critic. Each resolution is associated with a 'weight' (α<sub>4</sub>, α<sub>8</sub>, α<sub>16</sub>, α<sub>32</sub>), which indicate the focus on the corresponding image resolution at any given time during the training.
52
+
53
+ At the beginning of the training, α<sub>4</sub>=1, α<sub>8</sub>=0, α<sub>16</sub>=0, α<sub>32</sub>=0, with α<sub>4</sub>=0, α<sub>8</sub>=0, α<sub>16</sub>=0, α<sub>32</sub>=1 towards the end.