Hugging Face's logo

AbstractPhil
/
geo-beatrix

Image Classification
PyTorch
TensorBoard
Safetensors
cifar100
geometric-learning
fractal-encoding
in-training
no-attention
no-cross-entropy
Eval Results
Model card Files
xet
Metrics Community
AbstractPhil commited on
Commit
ad4b6b9
·
verified ·
1 Parent(s): 309bb35

Epoch 10: Test Acc 24.64%, α=0.4088

Browse files
Files changed (7) hide show
  1. README.md +176 -518
  2. best_model.pt +3 -0
  3. best_model.safetensors +3 -0
  4. config.json +28 -0
  5. latest_model.pt +3 -0
  6. latest_model.safetensors +3 -0
  7. training_log.txt +10 -0
README.md CHANGED
@@ -1,529 +1,187 @@
1
  ---
2
  license: mit
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  ---
4
 
5
- 10/9/2025 Retrain begins; lets see if she holds. This will be fully documented and uploaded over time like usual processes.
6
 
7
- I apologize for the pt, as this model and it's components will be directly integrated into the repo posthaste.
8
 
9
- It has defeated all beatrix/zana/pixie rivals out of sheer-force willpower as I modified a notebook over and over and over.
10
 
11
- cross-entropy is not in this training process at all.
12
 
13
- The full breakthrough will be available in an article very soon.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
14
 
 
 
 
 
 
 
 
15
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
16
  ```
17
- ================================================================================
18
- GEOMETRIC BASIN CLASSIFIER - CIFAR-100 [BIGGER EDITION]
19
- ================================================================================
20
-
21
- ✓ NO cross-entropy
22
- ✓ Classification via geometric formula basin checks
23
- ✓ BIGGER: Deeper backbone, more channels, residual connections
24
- ✓ Standard augmentations: Crop, Flip, ColorJitter, Rotation
25
- ✓ Better optimization: AdamW + Warmup + Cosine annealing
26
-
27
- Configuration:
28
- Batch size: 256 (BIGGER)
29
- Epochs: 200 (MORE)
30
- Base LR: 0.001
31
- Weight decay: 0.05
32
- Warmup epochs: 10
33
- Device: cuda
34
- PE levels: 20 (DEEPER hierarchy)
35
- PE features/level: 4 (RICHER)
36
- Total PE dim: 80
37
- Dropout: 0.1
38
-
39
- Loading CIFAR-100...
40
- Train: 50000 samples
41
- Test: 10000 samples
42
-
43
- Initializing BIGGER model...
44
- Total parameters: 45,161,489
45
- Trainable parameters: 45,161,489
46
- PE alpha (init): 0.1000
47
-
48
- Starting BIGGER training...
49
-
50
- Epoch 1: 100%|█████████████████████████████| 196/196 [00:14<00:00, 13.45it/s, loss=3.6482, acc=2.33%, α=0.1188, mix%=0.0, mixed=✗]
51
- Epoch 1/200 - Loss: 3.8428, Train Acc: 2.33%, α: 0.1188, LR: 0.000100, Mix: 0.0%
52
- Epoch 2: 100%|█████████████████████████████| 196/196 [00:14<00:00, 13.93it/s, loss=3.4946, acc=5.30%, α=0.1547, mix%=0.2, mixed=✗]
53
- Epoch 2/200 - Loss: 3.5909, Train Acc: 5.30%, α: 0.1547, LR: 0.000200, Mix: 0.0%
54
- Epoch 3: 100%|█████████████████████████████| 196/196 [00:14<00:00, 13.67it/s, loss=3.2373, acc=7.44%, α=0.2022, mix%=0.5, mixed=✗]
55
- Epoch 3/200 - Loss: 3.3676, Train Acc: 7.44%, α: 0.2022, LR: 0.000300, Mix: 0.0%
56
- Epoch 4: 100%|█��███████████████████████████| 196/196 [00:14<00:00, 13.96it/s, loss=3.0178, acc=9.43%, α=0.2554, mix%=0.8, mixed=✗]
57
- Epoch 4/200 - Loss: 3.1418, Train Acc: 9.43%, α: 0.2554, LR: 0.000400, Mix: 0.0%
58
- Epoch 5: 100%|████████████████████████████| 196/196 [00:14<00:00, 13.81it/s, loss=2.8231, acc=10.84%, α=0.3064, mix%=1.0, mixed=✗]
59
- Epoch 5/200 - Loss: 2.9193, Train Acc: 10.84%, α: 0.3064, LR: 0.000500, Mix: 1.0%
60
- → Test Acc: 12.60%
61
- ✓ Saved best model (acc: 12.60%)
62
- Epoch 6: 100%|████████████████████████████| 196/196 [00:14<00:00, 13.91it/s, loss=2.7102, acc=12.73%, α=0.3453, mix%=1.2, mixed=✗]
63
- Epoch 6/200 - Loss: 2.7181, Train Acc: 12.73%, α: 0.3453, LR: 0.000600, Mix: 1.5%
64
- Epoch 7: 100%|████████████████████████████| 196/196 [00:14<00:00, 13.72it/s, loss=2.5250, acc=15.00%, α=0.3728, mix%=1.5, mixed=✗]
65
- Epoch 7/200 - Loss: 2.5929, Train Acc: 15.00%, α: 0.3728, LR: 0.000700, Mix: 0.5%
66
- Epoch 8: 100%|████████████████████████████| 196/196 [00:14<00:00, 13.48it/s, loss=2.4662, acc=18.58%, α=0.3906, mix%=1.8, mixed=✗]
67
- Epoch 8/200 - Loss: 2.4405, Train Acc: 18.58%, α: 0.3906, LR: 0.000800, Mix: 3.1%
68
- Epoch 9: 100%|████████████████████████████| 196/196 [00:14<00:00, 13.49it/s, loss=2.4787, acc=22.45%, α=0.4047, mix%=2.0, mixed=✗]
69
- Epoch 9/200 - Loss: 2.4081, Train Acc: 22.45%, α: 0.4047, LR: 0.000900, Mix: 1.5%
70
- Epoch 10: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.93it/s, loss=2.4530, acc=25.12%, α=0.4127, mix%=2.2, mixed=✗]
71
- Epoch 10/200 - Loss: 2.3365, Train Acc: 25.12%, α: 0.4127, LR: 0.001000, Mix: 2.6%
72
- → Test Acc: 25.50%
73
- ✓ Saved best model (acc: 25.50%)
74
- Epoch 11: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.75it/s, loss=2.4797, acc=27.32%, α=0.4156, mix%=2.5, mixed=✗]
75
- Epoch 11/200 - Loss: 2.2918, Train Acc: 27.32%, α: 0.4156, LR: 0.001000, Mix: 1.5%
76
- Epoch 12: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.76it/s, loss=2.2486, acc=28.88%, α=0.4148, mix%=2.8, mixed=✗]
77
- Epoch 12/200 - Loss: 2.2308, Train Acc: 28.88%, α: 0.4148, LR: 0.001000, Mix: 2.6%
78
- Epoch 13: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.99it/s, loss=2.2080, acc=30.41%, α=0.4107, mix%=3.0, mixed=✗]
79
- Epoch 13/200 - Loss: 2.1588, Train Acc: 30.41%, α: 0.4107, LR: 0.001000, Mix: 4.6%
80
- Epoch 14: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.98it/s, loss=2.2585, acc=32.00%, α=0.4083, mix%=3.2, mixed=✗]
81
- Epoch 14/200 - Loss: 2.1440, Train Acc: 32.00%, α: 0.4083, LR: 0.000999, Mix: 3.1%
82
- Epoch 15: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.93it/s, loss=2.3343, acc=33.49%, α=0.4066, mix%=3.5, mixed=✗]
83
- Epoch 15/200 - Loss: 2.1110, Train Acc: 33.49%, α: 0.4066, LR: 0.000999, Mix: 3.1%
84
- → Test Acc: 35.19%
85
- ✓ Saved best model (acc: 35.19%)
86
- Epoch 16: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.97it/s, loss=2.2712, acc=34.92%, α=0.4041, mix%=3.8, mixed=✗]
87
- Epoch 16/200 - Loss: 2.0945, Train Acc: 34.92%, α: 0.4041, LR: 0.000998, Mix: 2.0%
88
- Epoch 17: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.42it/s, loss=2.1910, acc=36.01%, α=0.4033, mix%=4.0, mixed=✗]
89
- Epoch 17/200 - Loss: 2.0472, Train Acc: 36.01%, α: 0.4033, LR: 0.000998, Mix: 3.1%
90
- Epoch 18: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.62it/s, loss=2.0982, acc=37.21%, α=0.4027, mix%=4.2, mixed=✗]
91
- Epoch 18/200 - Loss: 2.0210, Train Acc: 37.21%, α: 0.4027, LR: 0.000997, Mix: 3.1%
92
- Epoch 19: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.91it/s, loss=2.2231, acc=38.23%, α=0.4012, mix%=4.5, mixed=✗]
93
- Epoch 19/200 - Loss: 1.9887, Train Acc: 38.23%, α: 0.4012, LR: 0.000996, Mix: 3.6%
94
- Epoch 20: 100%|███████████████████████████| 196/196 [00:13<00:00, 14.09it/s, loss=2.0983, acc=38.99%, α=0.4001, mix%=4.8, mixed=✗]
95
- Epoch 20/200 - Loss: 1.9251, Train Acc: 38.99%, α: 0.4001, LR: 0.000994, Mix: 6.1%
96
- → Test Acc: 38.03%
97
- ✓ Saved best model (acc: 38.03%)
98
- Epoch 21: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.80it/s, loss=1.9541, acc=40.42%, α=0.4013, mix%=5.0, mixed=✗]
99
- Epoch 21/200 - Loss: 1.9377, Train Acc: 40.42%, α: 0.4013, LR: 0.000993, Mix: 3.6%
100
- Epoch 22: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.49it/s, loss=2.0182, acc=41.20%, α=0.4011, mix%=5.2, mixed=✗]
101
- Epoch 22/200 - Loss: 1.8733, Train Acc: 41.20%, α: 0.4011, LR: 0.000992, Mix: 6.6%
102
- Epoch 23: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.83it/s, loss=1.9984, acc=42.49%, α=0.4020, mix%=5.5, mixed=✗]
103
- Epoch 23/200 - Loss: 1.8486, Train Acc: 42.49%, α: 0.4020, LR: 0.000990, Mix: 6.6%
104
- Epoch 24: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.80it/s, loss=1.9411, acc=43.20%, α=0.4023, mix%=5.8, mixed=✗]
105
- Epoch 24/200 - Loss: 1.8228, Train Acc: 43.20%, α: 0.4023, LR: 0.000988, Mix: 7.1%
106
- Epoch 25: 100%|███████████████████████████| 196/196 [00:13<00:00, 14.17it/s, loss=2.0109, acc=44.02%, α=0.4023, mix%=6.0, mixed=✗]
107
- Epoch 25/200 - Loss: 1.8007, Train Acc: 44.02%, α: 0.4023, LR: 0.000987, Mix: 7.1%
108
- → Test Acc: 43.50%
109
- ✓ Saved best model (acc: 43.50%)
110
- Epoch 26: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.91it/s, loss=1.9570, acc=45.09%, α=0.4009, mix%=6.2, mixed=✗]
111
- Epoch 26/200 - Loss: 1.7729, Train Acc: 45.09%, α: 0.4009, LR: 0.000985, Mix: 7.7%
112
- Epoch 27: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.86it/s, loss=1.9904, acc=45.93%, α=0.4039, mix%=6.5, mixed=✗]
113
- Epoch 27/200 - Loss: 1.7729, Train Acc: 45.93%, α: 0.4039, LR: 0.000983, Mix: 6.1%
114
- Epoch 28: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.71it/s, loss=1.8758, acc=47.09%, α=0.4050, mix%=6.8, mixed=✗]
115
- Epoch 28/200 - Loss: 1.7495, Train Acc: 47.09%, α: 0.4050, LR: 0.000980, Mix: 6.1%
116
- Epoch 29: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.97it/s, loss=1.8440, acc=46.74%, α=0.4048, mix%=7.0, mixed=✗]
117
- Epoch 29/200 - Loss: 1.6472, Train Acc: 46.74%, α: 0.4048, LR: 0.000978, Mix: 12.2%
118
- Epoch 30: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.94it/s, loss=1.7748, acc=48.32%, α=0.4065, mix%=7.2, mixed=✗]
119
- Epoch 30/200 - Loss: 1.7283, Train Acc: 48.32%, α: 0.4065, LR: 0.000976, Mix: 4.6%
120
- → Test Acc: 48.25%
121
- ✓ Saved best model (acc: 48.25%)
122
- Epoch 31: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.56it/s, loss=1.7850, acc=48.76%, α=0.4056, mix%=7.5, mixed=✗]
123
- Epoch 31/200 - Loss: 1.6609, Train Acc: 48.76%, α: 0.4056, LR: 0.000973, Mix: 7.7%
124
- Epoch 32: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.99it/s, loss=1.8050, acc=49.51%, α=0.4046, mix%=7.8, mixed=✗]
125
- Epoch 32/200 - Loss: 1.6460, Train Acc: 49.51%, α: 0.4046, LR: 0.000970, Mix: 7.1%
126
- Epoch 33: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.70it/s, loss=1.7683, acc=50.75%, α=0.4060, mix%=8.0, mixed=✗]
127
- Epoch 33/200 - Loss: 1.6336, Train Acc: 50.75%, α: 0.4060, LR: 0.000967, Mix: 6.1%
128
- Epoch 34: 100%|███████████████████████████| 196/196 [00:13<00:00, 14.10it/s, loss=0.4437, acc=51.31%, α=0.4061, mix%=8.2, mixed=✓]
129
- Epoch 34/200 - Loss: 1.5650, Train Acc: 51.31%, α: 0.4061, LR: 0.000964, Mix: 9.2%
130
- Epoch 35: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.88it/s, loss=1.7956, acc=51.79%, α=0.4092, mix%=8.5, mixed=✗]
131
- Epoch 35/200 - Loss: 1.5722, Train Acc: 51.79%, α: 0.4092, LR: 0.000961, Mix: 6.6%
132
- → Test Acc: 50.05%
133
- ✓ Saved best model (acc: 50.05%)
134
- Epoch 36: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.47it/s, loss=0.4191, acc=51.99%, α=0.4116, mix%=8.8, mixed=✓]
135
- Epoch 36/200 - Loss: 1.4906, Train Acc: 51.99%, α: 0.4116, LR: 0.000958, Mix: 11.2%
136
- Epoch 37: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.57it/s, loss=1.5406, acc=52.81%, α=0.4080, mix%=9.0, mixed=✗]
137
- Epoch 37/200 - Loss: 1.4740, Train Acc: 52.81%, α: 0.4080, LR: 0.000955, Mix: 10.2%
138
- Epoch 38: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.68it/s, loss=1.6052, acc=53.93%, α=0.4134, mix%=9.2, mixed=✗]
139
- Epoch 38/200 - Loss: 1.4730, Train Acc: 53.93%, α: 0.4134, LR: 0.000951, Mix: 7.7%
140
- Epoch 39: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.75it/s, loss=1.6939, acc=53.98%, α=0.4084, mix%=9.5, mixed=✗]
141
- Epoch 39/200 - Loss: 1.4151, Train Acc: 53.98%, α: 0.4084, LR: 0.000947, Mix: 11.2%
142
- Epoch 40: 100%|███████████████████████████| 196/196 [00:14<00:00, 13.60it/s, loss=1.5430, acc=54.53%, α=0.4109, mix%=9.8, mixed=✗]
143
- Epoch 40/200 - Loss: 1.4084, Train Acc: 54.53%, α: 0.4109, LR: 0.000944, Mix: 9.2%
144
- → Test Acc: 53.16%
145
- ✓ Saved best model (acc: 53.16%)
146
- Epoch 41: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.55it/s, loss=1.5753, acc=54.78%, α=0.4107, mix%=10.0, mixed=✗]
147
- Epoch 41/200 - Loss: 1.3655, Train Acc: 54.78%, α: 0.4107, LR: 0.000940, Mix: 11.2%
148
- Epoch 42: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.51it/s, loss=1.5854, acc=56.27%, α=0.4106, mix%=10.2, mixed=✗]
149
- Epoch 42/200 - Loss: 1.3686, Train Acc: 56.27%, α: 0.4106, LR: 0.000936, Mix: 8.7%
150
- Epoch 43: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.72it/s, loss=0.4049, acc=56.06%, α=0.4144, mix%=10.5, mixed=✓]
151
- Epoch 43/200 - Loss: 1.3302, Train Acc: 56.06%, α: 0.4144, LR: 0.000932, Mix: 10.7%
152
- Epoch 44: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.79it/s, loss=1.4748, acc=56.67%, α=0.4141, mix%=10.8, mixed=✗]
153
- Epoch 44/200 - Loss: 1.2810, Train Acc: 56.67%, α: 0.4141, LR: 0.000927, Mix: 13.3%
154
- Epoch 45: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.61it/s, loss=0.4018, acc=57.02%, α=0.4149, mix%=11.0, mixed=✓]
155
- Epoch 45/200 - Loss: 1.2782, Train Acc: 57.02%, α: 0.4149, LR: 0.000923, Mix: 12.8%
156
- → Test Acc: 55.44%
157
- ✓ Saved best model (acc: 55.44%)
158
- Epoch 46: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.76it/s, loss=1.2756, acc=58.43%, α=0.4125, mix%=11.2, mixed=✗]
159
- Epoch 46/200 - Loss: 1.2833, Train Acc: 58.43%, α: 0.4125, LR: 0.000919, Mix: 9.2%
160
- Epoch 47: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.45it/s, loss=1.4842, acc=58.44%, α=0.4151, mix%=11.5, mixed=✗]
161
- Epoch 47/200 - Loss: 1.2476, Train Acc: 58.44%, α: 0.4151, LR: 0.000914, Mix: 11.7%
162
- Epoch 48: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.43it/s, loss=1.3001, acc=59.41%, α=0.4180, mix%=11.8, mixed=✗]
163
- Epoch 48/200 - Loss: 1.2710, Train Acc: 59.41%, α: 0.4180, LR: 0.000909, Mix: 7.1%
164
- Epoch 49: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.57it/s, loss=0.4355, acc=59.85%, α=0.4196, mix%=12.0, mixed=✓]
165
- Epoch 49/200 - Loss: 1.2249, Train Acc: 59.85%, α: 0.4196, LR: 0.000905, Mix: 10.2%
166
- Epoch 50: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.71it/s, loss=1.3249, acc=60.09%, α=0.4204, mix%=12.2, mixed=✗]
167
- Epoch 50/200 - Loss: 1.1867, Train Acc: 60.09%, α: 0.4204, LR: 0.000900, Mix: 13.3%
168
- → Test Acc: 56.64%
169
- ✓ Saved best model (acc: 56.64%)
170
- Epoch 51: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.74it/s, loss=0.4623, acc=61.59%, α=0.4223, mix%=12.5, mixed=✓]
171
- Epoch 51/200 - Loss: 1.2174, Train Acc: 61.59%, α: 0.4223, LR: 0.000895, Mix: 8.2%
172
- Epoch 52: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.72it/s, loss=1.1845, acc=60.49%, α=0.4168, mix%=12.8, mixed=✗]
173
- Epoch 52/200 - Loss: 1.1496, Train Acc: 60.49%, α: 0.4168, LR: 0.000889, Mix: 14.8%
174
- Epoch 53: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.71it/s, loss=0.4386, acc=62.47%, α=0.4190, mix%=13.0, mixed=✓]
175
- Epoch 53/200 - Loss: 1.1679, Train Acc: 62.47%, α: 0.4190, LR: 0.000884, Mix: 10.7%
176
- Epoch 54: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.37it/s, loss=1.3397, acc=62.27%, α=0.4202, mix%=13.2, mixed=✗]
177
- Epoch 54/200 - Loss: 1.1369, Train Acc: 62.27%, α: 0.4202, LR: 0.000879, Mix: 12.8%
178
- Epoch 55: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.34it/s, loss=1.1730, acc=62.35%, α=0.4225, mix%=13.5, mixed=✗]
179
- Epoch 55/200 - Loss: 1.1168, Train Acc: 62.35%, α: 0.4225, LR: 0.000873, Mix: 14.3%
180
- → Test Acc: 57.25%
181
- ✓ Saved best model (acc: 57.25%)
182
- Epoch 56: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.50it/s, loss=0.4172, acc=63.03%, α=0.4222, mix%=13.8, mixed=✓]
183
- Epoch 56/200 - Loss: 1.1013, Train Acc: 63.03%, α: 0.4222, LR: 0.000868, Mix: 14.8%
184
- Epoch 57: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.19it/s, loss=1.3968, acc=63.75%, α=0.4280, mix%=14.0, mixed=✗]
185
- Epoch 57/200 - Loss: 1.0912, Train Acc: 63.75%, α: 0.4280, LR: 0.000862, Mix: 14.8%
186
- Epoch 58: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.51it/s, loss=1.1421, acc=65.04%, α=0.4256, mix%=14.2, mixed=✗]
187
- Epoch 58/200 - Loss: 1.1106, Train Acc: 65.04%, α: 0.4256, LR: 0.000856, Mix: 9.7%
188
- Epoch 59: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.56it/s, loss=1.2985, acc=65.26%, α=0.4268, mix%=14.5, mixed=✗]
189
- Epoch 59/200 - Loss: 1.0821, Train Acc: 65.26%, α: 0.4268, LR: 0.000851, Mix: 12.2%
190
- Epoch 60: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.52it/s, loss=1.3002, acc=65.13%, α=0.4296, mix%=14.8, mixed=✗]
191
- Epoch 60/200 - Loss: 1.0649, Train Acc: 65.13%, α: 0.4296, LR: 0.000845, Mix: 13.8%
192
- → Test Acc: 59.34%
193
- ✓ Saved best model (acc: 59.34%)
194
- Epoch 61: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.43it/s, loss=1.1267, acc=64.53%, α=0.4234, mix%=15.0, mixed=✗]
195
- Epoch 61/200 - Loss: 1.0217, Train Acc: 64.53%, α: 0.4234, LR: 0.000839, Mix: 18.4%
196
- Epoch 62: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.38it/s, loss=1.1919, acc=66.64%, α=0.4296, mix%=15.2, mixed=✗]
197
- Epoch 62/200 - Loss: 1.0563, Train Acc: 66.64%, α: 0.4296, LR: 0.000833, Mix: 11.7%
198
- Epoch 63: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.76it/s, loss=1.2270, acc=65.96%, α=0.4275, mix%=15.5, mixed=✗]
199
- Epoch 63/200 - Loss: 1.0371, Train Acc: 65.96%, α: 0.4275, LR: 0.000826, Mix: 14.3%
200
- Epoch 64: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.67it/s, loss=1.2329, acc=66.45%, α=0.4292, mix%=15.8, mixed=✗]
201
- Epoch 64/200 - Loss: 1.0204, Train Acc: 66.45%, α: 0.4292, LR: 0.000820, Mix: 15.3%
202
- Epoch 65: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.43it/s, loss=0.5662, acc=67.43%, α=0.4274, mix%=16.0, mixed=✓]
203
- Epoch 65/200 - Loss: 1.0083, Train Acc: 67.43%, α: 0.4274, LR: 0.000814, Mix: 14.3%
204
- → Test Acc: 60.67%
205
- ✓ Saved best model (acc: 60.67%)
206
- Epoch 66: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.12it/s, loss=0.5757, acc=67.39%, α=0.4264, mix%=16.2, mixed=✓]
207
- Epoch 66/200 - Loss: 1.0067, Train Acc: 67.39%, α: 0.4264, LR: 0.000807, Mix: 13.8%
208
- Epoch 67: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.42it/s, loss=1.1605, acc=68.30%, α=0.4297, mix%=16.5, mixed=✗]
209
- Epoch 67/200 - Loss: 1.0138, Train Acc: 68.30%, α: 0.4297, LR: 0.000801, Mix: 11.7%
210
- Epoch 68: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.50it/s, loss=1.2487, acc=68.17%, α=0.4278, mix%=16.8, mixed=✗]
211
- Epoch 68/200 - Loss: 0.9693, Train Acc: 68.17%, α: 0.4278, LR: 0.000794, Mix: 16.3%
212
- Epoch 69: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.59it/s, loss=1.1139, acc=68.73%, α=0.4333, mix%=17.0, mixed=✗]
213
- Epoch 69/200 - Loss: 0.9800, Train Acc: 68.73%, α: 0.4333, LR: 0.000787, Mix: 13.3%
214
- Epoch 70: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.62it/s, loss=1.0103, acc=68.76%, α=0.4306, mix%=17.2, mixed=✗]
215
- Epoch 70/200 - Loss: 0.9534, Train Acc: 68.76%, α: 0.4306, LR: 0.000780, Mix: 17.3%
216
- → Test Acc: 61.41%
217
- ✓ Saved best model (acc: 61.41%)
218
- Epoch 71: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.27it/s, loss=0.6507, acc=68.58%, α=0.4347, mix%=17.5, mixed=✓]
219
- Epoch 71/200 - Loss: 0.9361, Train Acc: 68.58%, α: 0.4347, LR: 0.000773, Mix: 18.4%
220
- Epoch 72: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.59it/s, loss=1.0173, acc=69.25%, α=0.4300, mix%=17.8, mixed=✗]
221
- Epoch 72/200 - Loss: 0.9316, Train Acc: 69.25%, α: 0.4300, LR: 0.000767, Mix: 17.9%
222
- Epoch 73: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.53it/s, loss=1.0945, acc=67.51%, α=0.4246, mix%=18.0, mixed=✗]
223
- Epoch 73/200 - Loss: 0.8929, Train Acc: 67.51%, α: 0.4246, LR: 0.000759, Mix: 25.5%
224
- Epoch 74: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.48it/s, loss=1.1568, acc=69.66%, α=0.4317, mix%=18.2, mixed=✗]
225
- Epoch 74/200 - Loss: 0.9084, Train Acc: 69.66%, α: 0.4317, LR: 0.000752, Mix: 18.9%
226
- Epoch 75: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.87it/s, loss=1.0822, acc=70.14%, α=0.4334, mix%=18.5, mixed=✗]
227
- Epoch 75/200 - Loss: 0.9150, Train Acc: 70.14%, α: 0.4334, LR: 0.000745, Mix: 16.3%
228
- → Test Acc: 62.67%
229
- ✓ Saved best model (acc: 62.67%)
230
- Epoch 76: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.46it/s, loss=1.0334, acc=70.50%, α=0.4358, mix%=18.8, mixed=✗]
231
- Epoch 76/200 - Loss: 0.9209, Train Acc: 70.50%, α: 0.4358, LR: 0.000738, Mix: 15.3%
232
- Epoch 77: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.35it/s, loss=1.0366, acc=70.56%, α=0.4314, mix%=19.0, mixed=✗]
233
- Epoch 77/200 - Loss: 0.8983, Train Acc: 70.56%, α: 0.4314, LR: 0.000731, Mix: 17.9%
234
- Epoch 78: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.67it/s, loss=0.9984, acc=70.46%, α=0.4305, mix%=19.2, mixed=✗]
235
- Epoch 78/200 - Loss: 0.8712, Train Acc: 70.46%, α: 0.4305, LR: 0.000723, Mix: 20.9%
236
- Epoch 79: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.31it/s, loss=0.4555, acc=71.96%, α=0.4324, mix%=19.5, mixed=✓]
237
- Epoch 79/200 - Loss: 0.8822, Train Acc: 71.96%, α: 0.4324, LR: 0.000716, Mix: 16.8%
238
- Epoch 80: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.23it/s, loss=0.6244, acc=70.51%, α=0.4287, mix%=19.8, mixed=✓]
239
- Epoch 80/200 - Loss: 0.8479, Train Acc: 70.51%, α: 0.4287, LR: 0.000708, Mix: 24.0%
240
- → Test Acc: 62.48%
241
- Epoch 81: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.54it/s, loss=0.9929, acc=71.25%, α=0.4338, mix%=20.0, mixed=✗]
242
- Epoch 81/200 - Loss: 0.8560, Train Acc: 71.25%, α: 0.4338, LR: 0.000701, Mix: 21.4%
243
- Epoch 82: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.49it/s, loss=0.9867, acc=71.32%, α=0.4353, mix%=20.2, mixed=✗]
244
- Epoch 82/200 - Loss: 0.8489, Train Acc: 71.32%, α: 0.4353, LR: 0.000693, Mix: 21.4%
245
- Epoch 83: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.74it/s, loss=0.4824, acc=71.52%, α=0.4321, mix%=20.5, mixed=✓]
246
- Epoch 83/200 - Loss: 0.8462, Train Acc: 71.52%, α: 0.4321, LR: 0.000686, Mix: 21.4%
247
- Epoch 84: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.48it/s, loss=0.9531, acc=71.94%, α=0.4310, mix%=20.8, mixed=✗]
248
- Epoch 84/200 - Loss: 0.8278, Train Acc: 71.94%, α: 0.4310, LR: 0.000678, Mix: 21.9%
249
- Epoch 85: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.45it/s, loss=1.0861, acc=72.22%, α=0.4340, mix%=21.0, mixed=✗]
250
- Epoch 85/200 - Loss: 0.8377, Train Acc: 72.22%, α: 0.4340, LR: 0.000670, Mix: 20.4%
251
- → Test Acc: 63.38%
252
- ✓ Saved best model (acc: 63.38%)
253
- Epoch 86: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.35it/s, loss=0.8573, acc=72.85%, α=0.4347, mix%=21.2, mixed=✗]
254
- Epoch 86/200 - Loss: 0.8301, Train Acc: 72.85%, α: 0.4347, LR: 0.000662, Mix: 20.4%
255
- Epoch 87: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.72it/s, loss=0.9000, acc=71.74%, α=0.4343, mix%=21.5, mixed=✗]
256
- Epoch 87/200 - Loss: 0.8001, Train Acc: 71.74%, α: 0.4343, LR: 0.000655, Mix: 25.5%
257
- Epoch 88: 100%|██████████████████████████| 196/196 [00:15<00:00, 13.04it/s, loss=0.5226, acc=74.53%, α=0.4368, mix%=21.8, mixed=✓]
258
- Epoch 88/200 - Loss: 0.8251, Train Acc: 74.53%, α: 0.4368, LR: 0.000647, Mix: 16.3%
259
- Epoch 89: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.86it/s, loss=0.9694, acc=72.86%, α=0.4314, mix%=22.0, mixed=✗]
260
- Epoch 89/200 - Loss: 0.7980, Train Acc: 72.86%, α: 0.4314, LR: 0.000639, Mix: 23.5%
261
- Epoch 90: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.75it/s, loss=0.5253, acc=71.85%, α=0.4319, mix%=22.2, mixed=✓]
262
- Epoch 90/200 - Loss: 0.7792, Train Acc: 71.85%, α: 0.4319, LR: 0.000631, Mix: 26.0%
263
- → Test Acc: 63.62%
264
- ✓ Saved best model (acc: 63.62%)
265
- Epoch 91: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.64it/s, loss=0.9219, acc=73.92%, α=0.4345, mix%=22.5, mixed=✗]
266
- Epoch 91/200 - Loss: 0.7913, Train Acc: 73.92%, α: 0.4345, LR: 0.000623, Mix: 21.4%
267
- Epoch 92: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.33it/s, loss=0.9374, acc=73.00%, α=0.4334, mix%=22.8, mixed=✗]
268
- Epoch 92/200 - Loss: 0.7734, Train Acc: 73.00%, α: 0.4334, LR: 0.000615, Mix: 25.0%
269
- Epoch 93: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.60it/s, loss=1.0261, acc=73.36%, α=0.4331, mix%=23.0, mixed=✗]
270
- Epoch 93/200 - Loss: 0.7849, Train Acc: 73.36%, α: 0.4331, LR: 0.000607, Mix: 23.0%
271
- Epoch 94: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.62it/s, loss=0.4277, acc=73.26%, α=0.4348, mix%=23.2, mixed=✓]
272
- Epoch 94/200 - Loss: 0.7758, Train Acc: 73.26%, α: 0.4348, LR: 0.000599, Mix: 22.4%
273
- Epoch 95: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.57it/s, loss=1.0474, acc=74.49%, α=0.4329, mix%=23.5, mixed=✗]
274
- Epoch 95/200 - Loss: 0.7762, Train Acc: 74.49%, α: 0.4329, LR: 0.000590, Mix: 19.4%
275
- → Test Acc: 64.54%
276
- ✓ Saved best model (acc: 64.54%)
277
- Epoch 96: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.43it/s, loss=0.8946, acc=72.56%, α=0.4336, mix%=23.8, mixed=✗]
278
- Epoch 96/200 - Loss: 0.7510, Train Acc: 72.56%, α: 0.4336, LR: 0.000582, Mix: 25.0%
279
- Epoch 97: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.99it/s, loss=0.8055, acc=71.07%, α=0.4321, mix%=24.0, mixed=✗]
280
- Epoch 97/200 - Loss: 0.7412, Train Acc: 71.07%, α: 0.4321, LR: 0.000574, Mix: 25.0%
281
- Epoch 98: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.65it/s, loss=0.8028, acc=69.38%, α=0.4337, mix%=24.2, mixed=✗]
282
- Epoch 98/200 - Loss: 0.7259, Train Acc: 69.38%, α: 0.4337, LR: 0.000566, Mix: 26.5%
283
- Epoch 99: 100%|██████████████████████████| 196/196 [00:14<00:00, 13.26it/s, loss=0.9556, acc=69.90%, α=0.4343, mix%=24.5, mixed=✗]
284
- Epoch 99/200 - Loss: 0.7294, Train Acc: 69.90%, α: 0.4343, LR: 0.000558, Mix: 25.5%
285
- Epoch 100: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.49it/s, loss=0.4095, acc=69.76%, α=0.4333, mix%=24.8, mixed=✓]
286
- Epoch 100/200 - Loss: 0.7254, Train Acc: 69.76%, α: 0.4333, LR: 0.000550, Mix: 25.5%
287
- → Test Acc: 64.20%
288
- Epoch 101: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.56it/s, loss=0.9485, acc=65.64%, α=0.4340, mix%=25.0, mixed=✗]
289
- Epoch 101/200 - Loss: 0.7096, Train Acc: 65.64%, α: 0.4340, LR: 0.000541, Mix: 28.6%
290
- Epoch 102: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.82it/s, loss=0.9626, acc=67.96%, α=0.4356, mix%=25.2, mixed=✗]
291
- Epoch 102/200 - Loss: 0.7141, Train Acc: 67.96%, α: 0.4356, LR: 0.000533, Mix: 27.0%
292
- Epoch 103: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.51it/s, loss=0.8757, acc=68.30%, α=0.4344, mix%=25.5, mixed=✗]
293
- Epoch 103/200 - Loss: 0.7159, Train Acc: 68.30%, α: 0.4344, LR: 0.000525, Mix: 25.0%
294
- Epoch 104: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.37it/s, loss=0.8474, acc=67.97%, α=0.4356, mix%=25.8, mixed=✗]
295
- Epoch 104/200 - Loss: 0.7058, Train Acc: 67.97%, α: 0.4356, LR: 0.000517, Mix: 27.0%
296
- Epoch 105: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.32it/s, loss=0.9419, acc=69.84%, α=0.4349, mix%=26.0, mixed=✗]
297
- Epoch 105/200 - Loss: 0.7079, Train Acc: 69.84%, α: 0.4349, LR: 0.000508, Mix: 25.0%
298
- → Test Acc: 65.12%
299
- ✓ Saved best model (acc: 65.12%)
300
- Epoch 106: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.49it/s, loss=0.7852, acc=65.72%, α=0.4342, mix%=26.2, mixed=✗]
301
- Epoch 106/200 - Loss: 0.6820, Train Acc: 65.72%, α: 0.4342, LR: 0.000500, Mix: 30.6%
302
- Epoch 107: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.50it/s, loss=0.9474, acc=65.57%, α=0.4350, mix%=26.5, mixed=✗]
303
- Epoch 107/200 - Loss: 0.6767, Train Acc: 65.57%, α: 0.4350, LR: 0.000492, Mix: 29.6%
304
- Epoch 108: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.57it/s, loss=0.8537, acc=69.08%, α=0.4362, mix%=26.8, mixed=✗]
305
- Epoch 108/200 - Loss: 0.6860, Train Acc: 69.08%, α: 0.4362, LR: 0.000483, Mix: 28.6%
306
- Epoch 109: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.33it/s, loss=0.4004, acc=70.18%, α=0.4355, mix%=27.0, mixed=✓]
307
- Epoch 109/200 - Loss: 0.6879, Train Acc: 70.18%, α: 0.4355, LR: 0.000475, Mix: 26.5%
308
- Epoch 110: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.11it/s, loss=0.4214, acc=65.72%, α=0.4348, mix%=27.3, mixed=✓]
309
- Epoch 110/200 - Loss: 0.6606, Train Acc: 65.72%, α: 0.4348, LR: 0.000467, Mix: 32.1%
310
- → Test Acc: 64.78%
311
- Epoch 111: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.42it/s, loss=0.8405, acc=68.89%, α=0.4371, mix%=27.5, mixed=✗]
312
- Epoch 111/200 - Loss: 0.6795, Train Acc: 68.89%, α: 0.4371, LR: 0.000459, Mix: 27.6%
313
- Epoch 112: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.65it/s, loss=0.7131, acc=72.60%, α=0.4405, mix%=27.8, mixed=✗]
314
- Epoch 112/200 - Loss: 0.6930, Train Acc: 72.60%, α: 0.4405, LR: 0.000450, Mix: 21.4%
315
- Epoch 113: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.38it/s, loss=0.5093, acc=68.10%, α=0.4375, mix%=28.0, mixed=✓]
316
- Epoch 113/200 - Loss: 0.6681, Train Acc: 68.10%, α: 0.4375, LR: 0.000442, Mix: 28.1%
317
- Epoch 114: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.16it/s, loss=0.9985, acc=71.80%, α=0.4400, mix%=28.2, mixed=✗]
318
- Epoch 114/200 - Loss: 0.6726, Train Acc: 71.80%, α: 0.4400, LR: 0.000434, Mix: 25.5%
319
- Epoch 115: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.47it/s, loss=0.8860, acc=65.66%, α=0.4389, mix%=28.5, mixed=✗]
320
- Epoch 115/200 - Loss: 0.6492, Train Acc: 65.66%, α: 0.4389, LR: 0.000426, Mix: 32.1%
321
- → Test Acc: 65.69%
322
- ✓ Saved best model (acc: 65.69%)
323
- Epoch 116: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.41it/s, loss=0.7919, acc=70.25%, α=0.4399, mix%=28.7, mixed=✗]
324
- Epoch 116/200 - Loss: 0.6638, Train Acc: 70.25%, α: 0.4399, LR: 0.000418, Mix: 25.5%
325
- Epoch 117: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.69it/s, loss=0.8560, acc=69.92%, α=0.4411, mix%=29.0, mixed=✗]
326
- Epoch 117/200 - Loss: 0.6549, Train Acc: 69.92%, α: 0.4411, LR: 0.000410, Mix: 26.5%
327
- Epoch 118: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.56it/s, loss=0.7234, acc=68.28%, α=0.4366, mix%=29.2, mixed=✗]
328
- Epoch 118/200 - Loss: 0.6461, Train Acc: 68.28%, α: 0.4366, LR: 0.000401, Mix: 30.1%
329
- Epoch 119: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.61it/s, loss=0.7587, acc=66.71%, α=0.4394, mix%=29.5, mixed=✗]
330
- Epoch 119/200 - Loss: 0.6348, Train Acc: 66.71%, α: 0.4394, LR: 0.000393, Mix: 30.1%
331
- Epoch 120: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.52it/s, loss=0.9973, acc=59.53%, α=0.4369, mix%=29.8, mixed=✗]
332
- Epoch 120/200 - Loss: 0.5955, Train Acc: 59.53%, α: 0.4369, LR: 0.000385, Mix: 42.9%
333
- → Test Acc: 65.22%
334
- Epoch 121: 100%|█████████████████████████| 196/196 [00:15<00:00, 12.74it/s, loss=0.7980, acc=69.17%, α=0.4404, mix%=30.0, mixed=✗]
335
- Epoch 121/200 - Loss: 0.6428, Train Acc: 69.17%, α: 0.4404, LR: 0.000377, Mix: 28.1%
336
- Epoch 122: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.33it/s, loss=0.7367, acc=70.00%, α=0.4415, mix%=30.2, mixed=✗]
337
- Epoch 122/200 - Loss: 0.6387, Train Acc: 70.00%, α: 0.4415, LR: 0.000369, Mix: 28.1%
338
- Epoch 123: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.45it/s, loss=0.7379, acc=70.39%, α=0.4401, mix%=30.5, mixed=✗]
339
- Epoch 123/200 - Loss: 0.6293, Train Acc: 70.39%, α: 0.4401, LR: 0.000361, Mix: 29.6%
340
- Epoch 124: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.77it/s, loss=0.8493, acc=64.89%, α=0.4410, mix%=30.8, mixed=✗]
341
- Epoch 124/200 - Loss: 0.6117, Train Acc: 64.89%, α: 0.4410, LR: 0.000353, Mix: 34.7%
342
- Epoch 125: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.72it/s, loss=0.7333, acc=70.49%, α=0.4410, mix%=31.0, mixed=✗]
343
- Epoch 125/200 - Loss: 0.6295, Train Acc: 70.49%, α: 0.4410, LR: 0.000345, Mix: 27.6%
344
- → Test Acc: 65.92%
345
- ✓ Saved best model (acc: 65.92%)
346
- Epoch 126: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.11it/s, loss=0.7841, acc=69.03%, α=0.4413, mix%=31.2, mixed=✗]
347
- Epoch 126/200 - Loss: 0.6200, Train Acc: 69.03%, α: 0.4413, LR: 0.000338, Mix: 30.1%
348
- Epoch 127: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.32it/s, loss=0.7585, acc=71.21%, α=0.4415, mix%=31.5, mixed=✗]
349
- Epoch 127/200 - Loss: 0.6227, Train Acc: 71.21%, α: 0.4415, LR: 0.000330, Mix: 28.6%
350
- Epoch 128: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.43it/s, loss=0.7801, acc=70.43%, α=0.4418, mix%=31.8, mixed=✗]
351
- Epoch 128/200 - Loss: 0.6183, Train Acc: 70.43%, α: 0.4418, LR: 0.000322, Mix: 28.6%
352
- Epoch 129: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.81it/s, loss=0.7585, acc=68.09%, α=0.4414, mix%=32.0, mixed=✗]
353
- Epoch 129/200 - Loss: 0.6047, Train Acc: 68.09%, α: 0.4414, LR: 0.000314, Mix: 32.7%
354
- Epoch 130: 100%|███████████���█████████████| 196/196 [00:14<00:00, 13.64it/s, loss=0.4042, acc=67.79%, α=0.4408, mix%=32.2, mixed=✓]
355
- Epoch 130/200 - Loss: 0.5965, Train Acc: 67.79%, α: 0.4408, LR: 0.000307, Mix: 34.7%
356
- → Test Acc: 65.33%
357
- Epoch 131: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.28it/s, loss=0.7260, acc=70.15%, α=0.4404, mix%=32.5, mixed=✗]
358
- Epoch 131/200 - Loss: 0.6065, Train Acc: 70.15%, α: 0.4404, LR: 0.000299, Mix: 29.6%
359
- Epoch 132: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.49it/s, loss=0.9200, acc=68.11%, α=0.4402, mix%=32.8, mixed=✗]
360
- Epoch 132/200 - Loss: 0.5905, Train Acc: 68.11%, α: 0.4402, LR: 0.000292, Mix: 34.7%
361
- Epoch 133: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.64it/s, loss=0.4226, acc=63.55%, α=0.4393, mix%=33.0, mixed=✓]
362
- Epoch 133/200 - Loss: 0.5837, Train Acc: 63.55%, α: 0.4393, LR: 0.000284, Mix: 36.7%
363
- Epoch 134: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.45it/s, loss=0.3982, acc=65.40%, α=0.4410, mix%=33.2, mixed=✓]
364
- Epoch 134/200 - Loss: 0.5781, Train Acc: 65.40%, α: 0.4410, LR: 0.000277, Mix: 37.2%
365
- Epoch 135: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.54it/s, loss=0.4414, acc=69.57%, α=0.4408, mix%=33.5, mixed=✓]
366
- Epoch 135/200 - Loss: 0.5922, Train Acc: 69.57%, α: 0.4408, LR: 0.000269, Mix: 32.1%
367
- → Test Acc: 66.18%
368
- ✓ Saved best model (acc: 66.18%)
369
- Epoch 136: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.63it/s, loss=0.9021, acc=70.21%, α=0.4426, mix%=33.8, mixed=✗]
370
- Epoch 136/200 - Loss: 0.5961, Train Acc: 70.21%, α: 0.4426, LR: 0.000262, Mix: 30.1%
371
- Epoch 137: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.55it/s, loss=0.7520, acc=71.70%, α=0.4421, mix%=34.0, mixed=✗]
372
- Epoch 137/200 - Loss: 0.5958, Train Acc: 71.70%, α: 0.4421, LR: 0.000255, Mix: 29.1%
373
- Epoch 138: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.40it/s, loss=0.7076, acc=66.68%, α=0.4416, mix%=34.2, mixed=✗]
374
- Epoch 138/200 - Loss: 0.5627, Train Acc: 66.68%, α: 0.4416, LR: 0.000248, Mix: 39.3%
375
- Epoch 139: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.27it/s, loss=0.4407, acc=70.19%, α=0.4419, mix%=34.5, mixed=✓]
376
- Epoch 139/200 - Loss: 0.5782, Train Acc: 70.19%, α: 0.4419, LR: 0.000241, Mix: 33.2%
377
- Epoch 140: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.82it/s, loss=0.3975, acc=65.36%, α=0.4410, mix%=34.8, mixed=✓]
378
- Epoch 140/200 - Loss: 0.5598, Train Acc: 65.36%, α: 0.4410, LR: 0.000233, Mix: 40.3%
379
- → Test Acc: 65.46%
380
- Epoch 141: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.50it/s, loss=0.6646, acc=70.35%, α=0.4425, mix%=35.0, mixed=✗]
381
- Epoch 141/200 - Loss: 0.5763, Train Acc: 70.35%, α: 0.4425, LR: 0.000227, Mix: 32.1%
382
- Epoch 142: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.57it/s, loss=0.6346, acc=68.42%, α=0.4427, mix%=35.2, mixed=✗]
383
- Epoch 142/200 - Loss: 0.5720, Train Acc: 68.42%, α: 0.4427, LR: 0.000220, Mix: 33.7%
384
- Epoch 143: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.49it/s, loss=0.7364, acc=67.42%, α=0.4432, mix%=35.5, mixed=✗]
385
- Epoch 143/200 - Loss: 0.5633, Train Acc: 67.42%, α: 0.4432, LR: 0.000213, Mix: 36.7%
386
- Epoch 144: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.44it/s, loss=0.4376, acc=69.91%, α=0.4416, mix%=35.8, mixed=✓]
387
- Epoch 144/200 - Loss: 0.5627, Train Acc: 69.91%, α: 0.4416, LR: 0.000206, Mix: 35.7%
388
- Epoch 145: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.61it/s, loss=0.3946, acc=68.17%, α=0.4412, mix%=36.0, mixed=✓]
389
- Epoch 145/200 - Loss: 0.5598, Train Acc: 68.17%, α: 0.4412, LR: 0.000199, Mix: 36.2%
390
- → Test Acc: 66.16%
391
- Epoch 146: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.69it/s, loss=0.7028, acc=68.07%, α=0.4426, mix%=36.2, mixed=✗]
392
- Epoch 146/200 - Loss: 0.5659, Train Acc: 68.07%, α: 0.4426, LR: 0.000193, Mix: 33.7%
393
- Epoch 147: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.34it/s, loss=0.6305, acc=64.16%, α=0.4414, mix%=36.5, mixed=✗]
394
- Epoch 147/200 - Loss: 0.5482, Train Acc: 64.16%, α: 0.4414, LR: 0.000186, Mix: 40.3%
395
- Epoch 148: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.46it/s, loss=0.3977, acc=68.39%, α=0.4413, mix%=36.8, mixed=✓]
396
- Epoch 148/200 - Loss: 0.5532, Train Acc: 68.39%, α: 0.4413, LR: 0.000180, Mix: 35.7%
397
- Epoch 149: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.76it/s, loss=0.4247, acc=67.12%, α=0.4427, mix%=37.0, mixed=✓]
398
- Epoch 149/200 - Loss: 0.5521, Train Acc: 67.12%, α: 0.4427, LR: 0.000174, Mix: 35.7%
399
- Epoch 150: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.57it/s, loss=0.6904, acc=63.94%, α=0.4419, mix%=37.2, mixed=✗]
400
- Epoch 150/200 - Loss: 0.5371, Train Acc: 63.94%, α: 0.4419, LR: 0.000167, Mix: 43.4%
401
- → Test Acc: 67.56%
402
- ✓ Saved best model (acc: 67.56%)
403
- Epoch 151: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.35it/s, loss=0.3889, acc=63.79%, α=0.4435, mix%=37.5, mixed=✓]
404
- Epoch 151/200 - Loss: 0.5333, Train Acc: 63.79%, α: 0.4435, LR: 0.000161, Mix: 43.4%
405
- Epoch 152: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.38it/s, loss=1.0364, acc=68.08%, α=0.4409, mix%=37.8, mixed=✗]
406
- Epoch 152/200 - Loss: 0.5398, Train Acc: 68.08%, α: 0.4409, LR: 0.000155, Mix: 40.8%
407
- Epoch 153: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.70it/s, loss=0.6776, acc=66.37%, α=0.4416, mix%=38.0, mixed=✗]
408
- Epoch 153/200 - Loss: 0.5380, Train Acc: 66.37%, α: 0.4416, LR: 0.000149, Mix: 40.3%
409
- Epoch 154: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.67it/s, loss=0.6557, acc=64.97%, α=0.4424, mix%=38.2, mixed=✗]
410
- Epoch 154/200 - Loss: 0.5361, Train Acc: 64.97%, α: 0.4424, LR: 0.000144, Mix: 41.3%
411
- Epoch 155: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.37it/s, loss=0.6404, acc=64.05%, α=0.4425, mix%=38.5, mixed=✗]
412
- Epoch 155/200 - Loss: 0.5354, Train Acc: 64.05%, α: 0.4425, LR: 0.000138, Mix: 40.8%
413
- → Test Acc: 67.22%
414
- Epoch 156: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.55it/s, loss=0.3864, acc=68.30%, α=0.4433, mix%=38.8, mixed=✓]
415
- Epoch 156/200 - Loss: 0.5328, Train Acc: 68.30%, α: 0.4433, LR: 0.000132, Mix: 41.3%
416
- Epoch 157: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.23it/s, loss=0.4384, acc=64.25%, α=0.4429, mix%=39.0, mixed=✓]
417
- Epoch 157/200 - Loss: 0.5212, Train Acc: 64.25%, α: 0.4429, LR: 0.000127, Mix: 45.4%
418
- Epoch 158: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.44it/s, loss=0.8105, acc=68.50%, α=0.4427, mix%=39.2, mixed=✗]
419
- Epoch 158/200 - Loss: 0.5329, Train Acc: 68.50%, α: 0.4427, LR: 0.000121, Mix: 39.8%
420
- Epoch 159: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.70it/s, loss=0.6363, acc=66.90%, α=0.4427, mix%=39.5, mixed=✗]
421
- Epoch 159/200 - Loss: 0.5228, Train Acc: 66.90%, α: 0.4427, LR: 0.000116, Mix: 42.9%
422
- Epoch 160: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.70it/s, loss=0.6388, acc=65.78%, α=0.4431, mix%=39.8, mixed=✗]
423
- Epoch 160/200 - Loss: 0.5316, Train Acc: 65.78%, α: 0.4431, LR: 0.000111, Mix: 40.3%
424
- → Test Acc: 66.58%
425
- Epoch 161: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.38it/s, loss=0.6915, acc=69.66%, α=0.4433, mix%=40.0, mixed=✗]
426
- Epoch 161/200 - Loss: 0.5333, Train Acc: 69.66%, α: 0.4433, LR: 0.000105, Mix: 38.3%
427
- Epoch 162: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.81it/s, loss=0.4277, acc=69.76%, α=0.4432, mix%=40.2, mixed=✓]
428
- Epoch 162/200 - Loss: 0.5285, Train Acc: 69.76%, α: 0.4432, LR: 0.000100, Mix: 39.3%
429
- Epoch 163: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.47it/s, loss=0.6230, acc=64.12%, α=0.4422, mix%=40.5, mixed=✗]
430
- Epoch 163/200 - Loss: 0.5223, Train Acc: 64.12%, α: 0.4422, LR: 0.000095, Mix: 42.9%
431
- Epoch 164: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.61it/s, loss=0.6684, acc=67.42%, α=0.4428, mix%=40.8, mixed=✗]
432
- Epoch 164/200 - Loss: 0.5263, Train Acc: 67.42%, α: 0.4428, LR: 0.000091, Mix: 39.8%
433
- Epoch 165: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.53it/s, loss=0.7079, acc=63.98%, α=0.4425, mix%=41.0, mixed=✗]
434
- Epoch 165/200 - Loss: 0.5217, Train Acc: 63.98%, α: 0.4425, LR: 0.000086, Mix: 43.4%
435
- → Test Acc: 67.19%
436
- Epoch 166: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.50it/s, loss=0.6059, acc=66.92%, α=0.4436, mix%=41.2, mixed=✗]
437
- Epoch 166/200 - Loss: 0.5263, Train Acc: 66.92%, α: 0.4436, LR: 0.000081, Mix: 39.3%
438
- Epoch 167: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.83it/s, loss=0.6809, acc=65.60%, α=0.4433, mix%=41.5, mixed=✗]
439
- Epoch 167/200 - Loss: 0.5243, Train Acc: 65.60%, α: 0.4433, LR: 0.000077, Mix: 40.3%
440
- Epoch 168: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.69it/s, loss=0.7620, acc=66.22%, α=0.4428, mix%=41.8, mixed=✗]
441
- Epoch 168/200 - Loss: 0.5204, Train Acc: 66.22%, α: 0.4428, LR: 0.000073, Mix: 41.8%
442
- Epoch 169: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.45it/s, loss=0.6654, acc=62.91%, α=0.4429, mix%=42.0, mixed=✗]
443
- Epoch 169/200 - Loss: 0.5119, Train Acc: 62.91%, α: 0.4429, LR: 0.000068, Mix: 45.4%
444
- Epoch 170: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.60it/s, loss=0.7948, acc=66.51%, α=0.4430, mix%=42.2, mixed=✗]
445
- Epoch 170/200 - Loss: 0.5151, Train Acc: 66.51%, α: 0.4430, LR: 0.000064, Mix: 42.9%
446
- → Test Acc: 67.62%
447
- ✓ Saved best model (acc: 67.62%)
448
- Epoch 171: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.54it/s, loss=0.3968, acc=68.10%, α=0.4434, mix%=42.5, mixed=✓]
449
- Epoch 171/200 - Loss: 0.5140, Train Acc: 68.10%, α: 0.4434, LR: 0.000060, Mix: 41.8%
450
- Epoch 172: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.64it/s, loss=0.4045, acc=67.01%, α=0.4429, mix%=42.8, mixed=✓]
451
- Epoch 172/200 - Loss: 0.5129, Train Acc: 67.01%, α: 0.4429, LR: 0.000056, Mix: 42.9%
452
- Epoch 173: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.84it/s, loss=0.6636, acc=65.63%, α=0.4430, mix%=43.0, mixed=✗]
453
- Epoch 173/200 - Loss: 0.5077, Train Acc: 65.63%, α: 0.4430, LR: 0.000053, Mix: 45.9%
454
- Epoch 174: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.90it/s, loss=0.5751, acc=68.30%, α=0.4432, mix%=43.2, mixed=✗]
455
- Epoch 174/200 - Loss: 0.5156, Train Acc: 68.30%, α: 0.4432, LR: 0.000049, Mix: 41.3%
456
- Epoch 175: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.49it/s, loss=0.6688, acc=67.72%, α=0.4429, mix%=43.5, mixed=✗]
457
- Epoch 175/200 - Loss: 0.5153, Train Acc: 67.72%, α: 0.4429, LR: 0.000045, Mix: 40.8%
458
- → Test Acc: 67.53%
459
- Epoch 176: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.81it/s, loss=0.6151, acc=68.57%, α=0.4433, mix%=43.8, mixed=✗]
460
- Epoch 176/200 - Loss: 0.5183, Train Acc: 68.57%, α: 0.4433, LR: 0.000042, Mix: 38.8%
461
- Epoch 177: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.82it/s, loss=0.6333, acc=62.46%, α=0.4430, mix%=44.0, mixed=✗]
462
- Epoch 177/200 - Loss: 0.4987, Train Acc: 62.46%, α: 0.4430, LR: 0.000039, Mix: 49.5%
463
- Epoch 178: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.92it/s, loss=0.4055, acc=63.61%, α=0.4429, mix%=44.2, mixed=✓]
464
- Epoch 178/200 - Loss: 0.5008, Train Acc: 63.61%, α: 0.4429, LR: 0.000036, Mix: 49.0%
465
- Epoch 179: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.76it/s, loss=0.6207, acc=68.88%, α=0.4431, mix%=44.5, mixed=✗]
466
- Epoch 179/200 - Loss: 0.5179, Train Acc: 68.88%, α: 0.4431, LR: 0.000033, Mix: 39.3%
467
- Epoch 180: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.91it/s, loss=0.6340, acc=62.84%, α=0.4430, mix%=44.8, mixed=✗]
468
- Epoch 180/200 - Loss: 0.5077, Train Acc: 62.84%, α: 0.4430, LR: 0.000030, Mix: 45.4%
469
- → Test Acc: 67.01%
470
- Epoch 181: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.68it/s, loss=0.3901, acc=62.25%, α=0.4430, mix%=45.0, mixed=✓]
471
- Epoch 181/200 - Loss: 0.4936, Train Acc: 62.25%, α: 0.4430, LR: 0.000027, Mix: 50.5%
472
- Epoch 182: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.56it/s, loss=0.6897, acc=64.24%, α=0.4432, mix%=45.2, mixed=✗]
473
- Epoch 182/200 - Loss: 0.5062, Train Acc: 64.24%, α: 0.4432, LR: 0.000024, Mix: 44.9%
474
- Epoch 183: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.51it/s, loss=0.5834, acc=59.99%, α=0.4430, mix%=45.5, mixed=✗]
475
- Epoch 183/200 - Loss: 0.4943, Train Acc: 59.99%, α: 0.4430, LR: 0.000022, Mix: 51.0%
476
- Epoch 184: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.28it/s, loss=0.4128, acc=69.69%, α=0.4431, mix%=45.8, mixed=✓]
477
- Epoch 184/200 - Loss: 0.5132, Train Acc: 69.69%, α: 0.4431, LR: 0.000020, Mix: 39.8%
478
- Epoch 185: 100%|██████████████████��██████| 196/196 [00:14<00:00, 13.59it/s, loss=0.4048, acc=60.68%, α=0.4428, mix%=46.0, mixed=✓]
479
- Epoch 185/200 - Loss: 0.4975, Train Acc: 60.68%, α: 0.4428, LR: 0.000017, Mix: 48.5%
480
- → Test Acc: 67.69%
481
- ✓ Saved best model (acc: 67.69%)
482
- Epoch 186: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.70it/s, loss=0.5764, acc=65.72%, α=0.4429, mix%=46.2, mixed=✗]
483
- Epoch 186/200 - Loss: 0.5064, Train Acc: 65.72%, α: 0.4429, LR: 0.000015, Mix: 44.4%
484
- Epoch 187: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.80it/s, loss=0.4034, acc=65.37%, α=0.4431, mix%=46.5, mixed=✓]
485
- Epoch 187/200 - Loss: 0.5065, Train Acc: 65.37%, α: 0.4431, LR: 0.000013, Mix: 43.9%
486
- Epoch 188: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.79it/s, loss=0.6861, acc=65.79%, α=0.4430, mix%=46.8, mixed=✗]
487
- Epoch 188/200 - Loss: 0.5074, Train Acc: 65.79%, α: 0.4430, LR: 0.000012, Mix: 43.9%
488
- Epoch 189: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.80it/s, loss=0.3997, acc=62.18%, α=0.4430, mix%=47.0, mixed=✓]
489
- Epoch 189/200 - Loss: 0.4990, Train Acc: 62.18%, α: 0.4430, LR: 0.000010, Mix: 48.5%
490
- Epoch 190: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.66it/s, loss=0.3986, acc=65.78%, α=0.4430, mix%=47.2, mixed=✓]
491
- Epoch 190/200 - Loss: 0.5000, Train Acc: 65.78%, α: 0.4430, LR: 0.000008, Mix: 45.9%
492
- → Test Acc: 66.78%
493
- Epoch 191: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.81it/s, loss=0.3911, acc=62.33%, α=0.4429, mix%=47.5, mixed=✓]
494
- Epoch 191/200 - Loss: 0.4921, Train Acc: 62.33%, α: 0.4429, LR: 0.000007, Mix: 51.5%
495
- Epoch 192: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.91it/s, loss=0.6440, acc=69.44%, α=0.4430, mix%=47.8, mixed=✗]
496
- Epoch 192/200 - Loss: 0.5131, Train Acc: 69.44%, α: 0.4430, LR: 0.000006, Mix: 40.3%
497
- Epoch 193: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.48it/s, loss=0.6184, acc=62.94%, α=0.4430, mix%=48.0, mixed=✗]
498
- Epoch 193/200 - Loss: 0.4941, Train Acc: 62.94%, α: 0.4430, LR: 0.000004, Mix: 50.5%
499
- Epoch 194: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.29it/s, loss=0.7793, acc=63.19%, α=0.4430, mix%=48.2, mixed=✗]
500
- Epoch 194/200 - Loss: 0.4951, Train Acc: 63.19%, α: 0.4430, LR: 0.000003, Mix: 49.5%
501
- Epoch 195: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.63it/s, loss=0.6024, acc=63.45%, α=0.4429, mix%=48.5, mixed=✗]
502
- Epoch 195/200 - Loss: 0.4998, Train Acc: 63.45%, α: 0.4429, LR: 0.000002, Mix: 48.0%
503
- → Test Acc: 67.20%
504
- Epoch 196: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.73it/s, loss=0.7275, acc=62.29%, α=0.4430, mix%=48.8, mixed=✗]
505
- Epoch 196/200 - Loss: 0.4960, Train Acc: 62.29%, α: 0.4430, LR: 0.000002, Mix: 49.5%
506
- Epoch 197: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.50it/s, loss=0.4051, acc=64.12%, α=0.4429, mix%=49.0, mixed=✓]
507
- Epoch 197/200 - Loss: 0.4959, Train Acc: 64.12%, α: 0.4429, LR: 0.000001, Mix: 49.0%
508
- Epoch 198: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.54it/s, loss=0.4083, acc=62.07%, α=0.4429, mix%=49.2, mixed=✓]
509
- Epoch 198/200 - Loss: 0.5031, Train Acc: 62.07%, α: 0.4429, LR: 0.000001, Mix: 45.9%
510
- Epoch 199: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.58it/s, loss=0.6686, acc=61.61%, α=0.4429, mix%=49.5, mixed=✗]
511
- Epoch 199/200 - Loss: 0.4984, Train Acc: 61.61%, α: 0.4429, LR: 0.000000, Mix: 48.5%
512
- Epoch 200: 100%|█████████████████████████| 196/196 [00:14<00:00, 13.89it/s, loss=0.3969, acc=58.55%, α=0.4429, mix%=49.8, mixed=✓]
513
- Epoch 200/200 - Loss: 0.4828, Train Acc: 58.55%, α: 0.4429, LR: 0.000000, Mix: 56.6%
514
- → Test Acc: 67.67%
515
-
516
- ================================================================================
517
- TRAINING COMPLETE
518
- ================================================================================
519
-
520
- Final Test Accuracy: 67.69%
521
- Best Test Accuracy: 67.69%
522
- Final PE Alpha: 0.4428
523
- Total Parameters: 45,161,489
524
-
525
- ✓ BIGGER: Deeper backbone, more channels, residual connections
526
- ✓ Standard augmentations + AlphaMix curriculum
527
- ✓ Pure geometric basin classification - NO cross-entropy
528
-
529
- ```
 
1
  ---
2
  license: mit
3
+ tags:
4
+ - image-classification
5
+ - cifar100
6
+ - geometric-learning
7
+ - fractal-encoding
8
+ - in-training
9
+ - no-attention
10
+ - no-cross-entropy
11
+ datasets:
12
+ - cifar100
13
+ metrics:
14
+ - accuracy
15
+ library_name: pytorch
16
+ pipeline_tag: image-classification
17
+ model-index:
18
+ - name: geo-beatrix
19
+ results:
20
+ - task:
21
+ type: image-classification
22
+ name: Image Classification
23
+ dataset:
24
+ name: CIFAR-100
25
+ type: cifar100
26
+ metrics:
27
+ - type: accuracy
28
+ value: 24.64
29
+ name: Test Accuracy
30
+ verified: false
31
  ---
32
 
33
+ # geo-beatrix
34
 
35
+ **Geometric Basin Classification for CIFAR-100**
36
 
37
+ 🚧 **Training in Progress** 🚧
38
 
39
+ Current Status: Epoch 10/200
40
 
41
+ ---
42
+
43
+ ## Current Performance
44
+
45
+ | Metric | Value |
46
+ |--------|-------|
47
+ | **Best Test Accuracy** | **24.64%** |
48
+ | **Best Epoch** | 10 |
49
+ | **Current Train Accuracy** | 23.05% |
50
+ | **Current Test Accuracy** | 24.64% |
51
+ | **Current α (Cantor param)** | 0.4088 |
52
+ | **Total Parameters** | 45,161,489 |
53
+ | **Training Time** | 0:02:45 |
54
+
55
+ ### Comparison to State-of-the-Art
56
+
57
+ | Model | Accuracy | Status |
58
+ |-------|----------|--------|
59
+ | **geo-beatrix (this model)** | **24.64%** | 🔄 Training |
60
+ | vit-beatrix-dualstream | 66.0% | Vision Transformer + Cross-Entropy |
61
+ | CLIP ViT-L/14 (zero-shot) | ~63-65% | 400M image-text pairs |
62
+ | CLIP ViT-B/32 (zero-shot) | ~63.5% | Vision Transformer |
63
+
64
+ 🎯 **Current target**: Beat vit-beatrix (66.0%) - Currently -41.36%
65
+
66
+ ---
67
+
68
+ ## Architecture
69
 
70
+ - **Base**: ResNet-style with residual blocks
71
+ - **Channels**: 64 → 128 → 256 → 512 → 1024
72
+ - **Positional Encoding**: Devil's Staircase (Cantor function, 1883)
73
+ - **PE Levels**: 20
74
+ - **PE Features/Level**: 4
75
+ - **Classification**: Geometric Basin Compatibility (NO cross-entropy)
76
+ - **Attention Mechanisms**: NONE
77
 
78
+ ---
79
+
80
+ ## Training Configuration
81
+
82
+ ```json
83
+ {
84
+ "model_name": "geo-beatrix",
85
+ "model_type": "geometric_basin_classifier",
86
+ "num_classes": 100,
87
+ "batch_size": 256,
88
+ "num_epochs": 200,
89
+ "base_learning_rate": 0.001,
90
+ "weight_decay": 0.05,
91
+ "warmup_epochs": 10,
92
+ "pe_levels": 20,
93
+ "pe_features_per_level": 4,
94
+ "dropout": 0.1,
95
+ "alphamix": {
96
+ "enabled": true,
97
+ "range": [
98
+ 0.3,
99
+ 0.7
100
+ ],
101
+ "spatial_ratio": 0.25,
102
+ "curriculum_start": 0.0,
103
+ "curriculum_end": 0.5
104
+ },
105
+ "architecture": "ResNet-style with Devil's Staircase PE",
106
+ "loss_function": "Geometric Basin Compatibility",
107
+ "cross_entropy": false,
108
+ "attention_mechanisms": false,
109
+ "timestamp": "20251009_213912"
110
+ }
111
+ ```
112
+
113
+ ---
114
+
115
+ ## Training History
116
+
117
+ ### Training Progress
118
+
119
+ ```
120
+ Epochs: 10/200 (█ ) 5.0%
121
+ Best Acc: 24.64%
122
+ Current α: 0.4088 → Target: 0.50 (triadic equilibrium)
123
+ Status: 🔄 Training In Progress
124
  ```
125
+
126
+ ### Best Checkpoint
127
+ - Epoch: 10
128
+ - Train Acc: 23.05%
129
+ - Test Acc: 24.64%
130
+ - Alpha: 0.4088
131
+ - Loss: 2.3491
132
+
133
+ ### Latest 5 Epochs
134
+
135
+ - **Epoch 6**: Train 10.90%, Test 0.00%, α=0.3409, Loss=2.7317
136
+ - **Epoch 7**: Train 13.44%, Test 0.00%, α=0.3685, Loss=2.5983
137
+ - **Epoch 8**: Train 16.74%, Test 0.00%, α=0.3884, Loss=2.4982
138
+ - **Epoch 9**: Train 20.61%, Test 0.00%, α=0.4016, Loss=2.4181
139
+ - **Epoch 10**: Train 23.05%, Test 24.64%, α=0.4088, Loss=2.3491
140
+
141
+ ### Training Milestones
142
+ - 📊 **α ≥ 0.40** reached at epoch 9
143
+
144
+
145
+ ---
146
+
147
+ ## Usage
148
+
149
+ ```python
150
+ from huggingface_hub import hf_hub_download
151
+ import torch
152
+
153
+ # Option 1: Download safetensors (recommended - faster, safer)
154
+ from safetensors.torch import load_file
155
+ model_path = hf_hub_download(
156
+ repo_id="AbstractPhil/geo-beatrix",
157
+ filename="best_model.safetensors"
158
+ )
159
+ state_dict = load_file(model_path)
160
+ # model.load_state_dict(state_dict)
161
+
162
+ # Option 2: Download PyTorch checkpoint (includes optimizer state)
163
+ checkpoint_path = hf_hub_download(
164
+ repo_id="AbstractPhil/geo-beatrix",
165
+ filename="best_model.pt"
166
+ )
167
+ checkpoint = torch.load(checkpoint_path)
168
+ # model.load_state_dict(checkpoint['model_state_dict'])
169
+ ```
170
+
171
+ ---
172
+
173
+ ## Innovation
174
+
175
+ ✅ **NO attention mechanisms**
176
+ ✅ **NO cross-entropy loss**
177
+ ✅ **Fractal positional encoding** (Cantor function from 1883)
178
+ ✅ **Geometric compatibility classification**
179
+ ✅ **Ancient convolutions** (ResNet-style from 1990s)
180
+
181
+ ---
182
+
183
+ *This model is currently training. Check back for updates!*
184
+
185
+ **Repository**: https://huggingface.co/AbstractPhil/geo-beatrix
186
+ **Author**: AbstractPhil
187
+ **Framework**: PyTorch
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
best_model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:36f70a680da84a9a69a7972b267b1e6bafaefb181c198cc4a3ec59efbe85f208
3
+ size 542124404
best_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7203b9d8d18c2e97677a86b42c9395ee205a195a2e2ea5af0d898fa6422dfe33
3
+ size 180735732
config.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_name": "geo-beatrix",
3
+ "model_type": "geometric_basin_classifier",
4
+ "num_classes": 100,
5
+ "batch_size": 256,
6
+ "num_epochs": 200,
7
+ "base_learning_rate": 0.001,
8
+ "weight_decay": 0.05,
9
+ "warmup_epochs": 10,
10
+ "pe_levels": 20,
11
+ "pe_features_per_level": 4,
12
+ "dropout": 0.1,
13
+ "alphamix": {
14
+ "enabled": true,
15
+ "range": [
16
+ 0.3,
17
+ 0.7
18
+ ],
19
+ "spatial_ratio": 0.25,
20
+ "curriculum_start": 0.0,
21
+ "curriculum_end": 0.5
22
+ },
23
+ "architecture": "ResNet-style with Devil's Staircase PE",
24
+ "loss_function": "Geometric Basin Compatibility",
25
+ "cross_entropy": false,
26
+ "attention_mechanisms": false,
27
+ "timestamp": "20251009_213912"
28
+ }
latest_model.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:214517eb8abab75161f14fb955d7e23a6006f47c8adfe228ed2d3c29261c2e95
3
+ size 542125126
latest_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:845cc2a061191c4b2957327f85f7f27a85dcb9997fc8362eb997cdb55d42aa85
3
+ size 180735732
training_log.txt ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ Epoch 1/200 - Loss: 3.8604, Train Acc: 1.90%, Test Acc: 0.00%, α: 0.1190, LR: 0.000100, Mix: 0.0%
2
+ Epoch 2/200 - Loss: 3.6229, Train Acc: 4.05%, Test Acc: 0.00%, α: 0.1549, LR: 0.000200, Mix: 0.0%
3
+ Epoch 3/200 - Loss: 3.3932, Train Acc: 5.60%, Test Acc: 0.00%, α: 0.2029, LR: 0.000300, Mix: 0.5%
4
+ Epoch 4/200 - Loss: 3.1860, Train Acc: 7.22%, Test Acc: 0.00%, α: 0.2550, LR: 0.000400, Mix: 0.0%
5
+ Epoch 5/200 - Loss: 2.9280, Train Acc: 8.95%, Test Acc: 10.24%, α: 0.3024, LR: 0.000500, Mix: 2.0%
6
+ Epoch 6/200 - Loss: 2.7317, Train Acc: 10.90%, Test Acc: 0.00%, α: 0.3409, LR: 0.000600, Mix: 2.0%
7
+ Epoch 7/200 - Loss: 2.5983, Train Acc: 13.44%, Test Acc: 0.00%, α: 0.3685, LR: 0.000700, Mix: 1.5%
8
+ Epoch 8/200 - Loss: 2.4982, Train Acc: 16.74%, Test Acc: 0.00%, α: 0.3884, LR: 0.000800, Mix: 1.5%
9
+ Epoch 9/200 - Loss: 2.4181, Train Acc: 20.61%, Test Acc: 0.00%, α: 0.4016, LR: 0.000900, Mix: 2.0%
10
+ Epoch 10/200 - Loss: 2.3491, Train Acc: 23.05%, Test Acc: 24.64%, α: 0.4088, LR: 0.001000, Mix: 3.1%