File size: 959 Bytes
b23f673
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
{
  "type": "submodel",
  "name": "gradfield-cnn-final",
  "version": "3.0.0-finetuned",
  "framework": "pytorch",
  "arch": "CompactGradientNet",
  "model_parameters": {
    "depth": 4,
    "base_filters": 32,
    "dropout": 0.15,
    "embedding_dim": 128
  },
  "input": {
    "type": "luminance",
    "channels": 1,
    "size": 224,
    "preprocessing": "BT.709 luminance conversion"
  },
  "output": {
    "type": "binary_logit",
    "labels": {
      "0": "real",
      "1": "fake"
    }
  },
  "best_auroc": 0.8453642337807606,
  "finetune_config": {
    "depth": 4,
    "base_filters": 32,
    "embedding_dim": 128,
    "head_lr": 0.0001,
    "head_epochs": 5,
    "full_lr": 3e-05,
    "full_epochs": 20,
    "warmup_epochs": 2,
    "min_lr": 1e-07,
    "weight_decay": 0.001,
    "dropout": 0.15,
    "batch_size": 64,
    "patience": 5,
    "mixup_alpha": 0.2,
    "grad_clip": 1.0,
    "checkpoint_name": "gradient_field_cnn_v3_finetuned.pth"
  }
}