Hugging Face's logo

WillNuyen228
/
auramask-v2-oilify

Image-to-Image
Keras
LiteRT
adversarial
aesthetic
quality
filter
Model card Files
xet
 Community
WillNuyen228 commited on
Commit
2f406be
·
verified ·
1 Parent(s): 6b21b02

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +118 -9
README.md CHANGED
@@ -1,15 +1,124 @@
1
-
2
  ---
 
 
 
 
 
 
3
  library_name: keras
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4
  ---
 
 
 
 
 
 
5
 
6
- This model has been uploaded using the Keras library and can be used with JAX,
7
- TensorFlow, and PyTorch backends.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8
 
9
- This model card has been generated automatically and should be completed by the
10
- model author.
11
- See [Model Cards documentation](https://huggingface.co/docs/hub/model-cards) for
12
- more information.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
13
 
14
- For more details about the model architecture, check out
15
- [config.json](./config.json).A plot of the model can be found [here](./assets/summary_plot.png).
 
 
1
  ---
2
+ base_model:
3
+ - vnet
4
+ - logasja/ArcFace
5
+ - logasja/VGGFace
6
+ datasets:
7
+ - logasja/FDF
8
  library_name: keras
9
+ license: gpl-3.0
10
+ metrics:
11
+ - TopIQ-FR
12
+ - ArcFace Cosine Distance
13
+ - VGGFace2 Cosine Distance
14
+ pipeline_tag: image-to-image
15
+ tags:
16
+ - adversarial
17
+ - aesthetic
18
+ - quality
19
+ - filter
20
+ widget:
21
+ - text: input
22
+ output:
23
+ url: ./assets/input.png
24
+ - text: target
25
+ output:
26
+ url: ./assets/target.png
27
+ - text: output
28
+ output:
29
+ url: ./assets/output.png
30
  ---
31
+ <Gallery />
32
+
33
+ Training logs [here](https://wandb.ai/spuds/auramask/runs/i9gai52u)
34
+
35
+ # Model Description
36
+ Looks surprisingly legit
37
 
38
+ ```json
39
+ {
40
+ "D": [
41
+ 1,
42
+ 1,
43
+ 1,
44
+ 1,
45
+ 1
46
+ ],
47
+ "E": [
48
+ 1,
49
+ 1,
50
+ 2,
51
+ 3,
52
+ 5
53
+ ],
54
+ "activation": "relu",
55
+ "batch_norm": false,
56
+ "filter_num": [
57
+ 16,
58
+ 32,
59
+ 64,
60
+ 128,
61
+ 128
62
+ ],
63
+ "kernel_reg": "l2",
64
+ "n_labels": 3,
65
+ "output_activation": null
66
+ }
67
+ ```
68
 
69
+ ```json
70
+ {
71
+ "alpha": 0.0001,
72
+ "batch": 16,
73
+ "epochs": 50,
74
+ "epsilon": 0.03,
75
+ "input": "(256, 256)",
76
+ "losses": {
77
+ "FEAT_ArcFace": {
78
+ "d": "cosine_similarity",
79
+ "f": "ArcFace",
80
+ "name": "FEAT_ArcFace",
81
+ "reduction": "sum_over_batch_size",
82
+ "threshold": 0.68,
83
+ "weight": 0.1
84
+ },
85
+ "TopIQ": {
86
+ "full_ref": true,
87
+ "lower_better": false,
88
+ "name": "TopIQ",
89
+ "reduction": "sum_over_batch_size",
90
+ "score_range": "~0, ~1",
91
+ "weight": 0.9
92
+ },
93
+ "mean_squared_error": {
94
+ "name": "mean_squared_error",
95
+ "reduction": "sum_over_batch_size",
96
+ "weight": 0.1
97
+ }
98
+ },
99
+ "mixed_precision": false,
100
+ "optimizer": {
101
+ "amsgrad": false,
102
+ "beta_1": 0.9,
103
+ "beta_2": 0.999,
104
+ "clipnorm": 1,
105
+ "clipvalue": null,
106
+ "ema_momentum": 0.99,
107
+ "ema_overwrite_frequency": null,
108
+ "epsilon": 1e-07,
109
+ "global_clipnorm": null,
110
+ "gradient_accumulation_steps": null,
111
+ "learning_rate": 9.999999747378752e-05,
112
+ "loss_scale_factor": null,
113
+ "name": "adam",
114
+ "use_ema": false,
115
+ "weight_decay": null
116
+ },
117
+ "seed": "AEDDONQTKBCA",
118
+ "testing": 0.1,
119
+ "training": 0.9
120
+ }
121
+ ```
122
 
123
+ ## Model Architecture Plot
124
+ ![](./assets/summary_plot.png)