Hugging Face's logo

H1merka
/
TIGAS

Image Classification
PyTorch
English
tigas
computer-vision
ai-detection
deepfake-detection
image-quality
Eval Results
Model card Files
xet
Community
TIGAS
File size: 1,638 Bytes 
aac32b4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
 
{
  "model_type": "tigas",
  "architectures": ["TIGASModel"],
  "task_type": "image-classification",
  "framework": "pytorch",
  "model_config": {
    "img_size": 256,
    "in_channels": 3,
    "feature_dim": 256,
    "base_channels": 32,
    "num_scales": 4,
    "fast_mode": false
  },
  "training_config": {
    "epochs_trained": 3,
    "batch_size": 8,
    "learning_rate": 0.0001,
    "optimizer": "adamw",
    "scheduler": "cosine",
    "mixed_precision": true,
    "warmup_epochs": 5
  },
  "dataset_info": {
    "train_samples": 128776,
    "val_samples": 14167,
    "test_samples": 14126,
    "total_samples": 157069,
    "real_ratio": 0.458,
    "fake_ratio": 0.542
  },
  "metrics": {
    "best_val_loss": 0.3079,
    "best_val_accuracy": 0.6555,
    "final_train_loss": 0.3506
  },
  "input_spec": {
    "type": "image",
    "channels": 3,
    "height": 256,
    "width": 256,
    "normalization": {
      "mean": [0.5, 0.5, 0.5],
      "std": [0.5, 0.5, 0.5],
      "range": [-1, 1]
    }
  },
  "output_spec": {
    "type": "score",
    "range": [0, 1],
    "interpretation": {
      "1.0": "real/natural image",
      "0.0": "fake/generated image"
    }
  },
  "checkpoint_info": {
    "format": "pytorch",
    "keys": [
      "model_state_dict",
      "optimizer_state_dict", 
      "scheduler_state_dict",
      "scaler_state_dict",
      "epoch",
      "global_step",
      "best_val_loss",
      "train_history",
      "val_history"
    ]
  },
  "version": "0.1.0",
  "library_name": "tigas",
  "github_repo": "https://github.com/H1merka/TIGAS"
}