File size: 3,138 Bytes
4f58e42
aa0bed2
206438f
aa0bed2
 
 
 
 
4f58e42
 
 
 
 
aa0bed2
 
4f58e42
aa0bed2
 
4f58e42
 
 
 
 
 
 
aa0bed2
4f58e42
 
 
 
 
 
 
 
 
 
 
aa0bed2
 
4f58e42
 
 
 
 
 
 
aa0bed2
 
 
 
4f58e42
 
 
 
 
 
 
 
 
aa0bed2
 
 
4f58e42
 
 
 
 
 
 
 
 
 
 
aa0bed2
4f58e42
 
 
aa0bed2
4f58e42
 
 
aa0bed2
4f58e42
 
 
aa0bed2
4f58e42
 
aa0bed2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4f58e42
 
 
 
 
aa0bed2
 
4f58e42
 
 
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
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
{
  "methodology": "Real PyTorch 20-epoch mini-training with fault injection",
  "torch_version": "2.5.1+cpu",
  "models": [
    "SimpleCNN (~50K params)",
    "SimpleMLP (~20K params)"
  ],
  "training_approach": "Real forward+backward passes on random CIFAR-10 style data, cached per (task_id, seed)",
  "results": [
    {
      "task": "task_001",
      "fault": "exploding_gradients",
      "checks": {
        "gradient_instability_detected": true,
        "loss_shows_instability": true,
        "max_gradient_norm": 111.8,
        "max_loss": 43.27,
        "real_pytorch_training": true
      },
      "pass": true
    },
    {
      "task": "task_002",
      "fault": "vanishing_gradients",
      "checks": {
        "vanishing_detected": true,
        "min_gradient_norm": 0.0,
        "real_pytorch_gradients": true
      },
      "pass": true
    },
    {
      "task": "task_003",
      "fault": "data_leakage",
      "checks": {
        "class_overlap_above_0.5": true,
        "class_overlap_score": 0.83,
        "real_training_runs": true,
        "has_confusion_matrix": true
      },
      "pass": true
    },
    {
      "task": "task_004",
      "fault": "overfitting",
      "checks": {
        "real_training_runs": true,
        "clean_data": true,
        "final_train_loss": 0.1017,
        "final_val_loss": 2.6519
      },
      "pass": true
    },
    {
      "task": "task_005",
      "fault": "batchnorm_eval_mode",
      "checks": {
        "all_layers_in_eval_mode": true,
        "no_layer_is_exploding": true,
        "real_training_runs": true,
        "real_model_eval_mode": true,
        "red_herring_spike_layer": "conv1"
      },
      "pass": true
    },
    {
      "task": "task_006",
      "fault": "code_bug",
      "checks": {
        "variants_tested": 4,
        "variant_results": {
          "eval_mode": {
            "correct_fix_accepted": true,
            "wrong_fix_rejected": true
          },
          "detach_loss": {
            "correct_fix_accepted": true,
            "wrong_fix_rejected": true
          },
          "zero_grad_missing": {
            "correct_fix_accepted": true,
            "wrong_fix_rejected": true
          },
          "inplace_relu": {
            "correct_fix_accepted": true,
            "wrong_fix_rejected": true
          }
        },
        "fix_validation_pipeline": "normalize -> tokenize -> semantic -> AST"
      },
      "pass": true
    },
    {
      "task": "task_007",
      "fault": "scheduler_misconfigured",
      "checks": {
        "real_training_runs": true,
        "scheduler_gamma": 0.0001,
        "scheduler_step_size": 2,
        "final_loss": 2.5911
      },
      "pass": true
    },
    {
      "task": "architecture",
      "fault": "dual_model_support",
      "checks": {
        "cnn_output_shape": [
          4,
          10
        ],
        "mlp_output_shape": [
          4,
          10
        ],
        "cnn_params": 66890,
        "mlp_params": 411658,
        "both_produce_10_classes": true
      },
      "pass": true
    }
  ],
  "summary": {
    "total": 8,
    "passed": 8,
    "failed": 0
  }
}