AbstractPhil commited on
Commit
7437b02
·
verified ·
1 Parent(s): f20d73b

Upload weights - GeoFractalDavid-Basin-k50 - Run 20251015_222757 - Acc 80.36%

Browse files
weights/GeoFractalDavid-Basin-k50/20251015_222757/README.md CHANGED
@@ -23,7 +23,7 @@ model-index:
23
  type: imagenet-1k
24
  metrics:
25
  - type: accuracy
26
- value: 79.01
27
  name: Validation Accuracy
28
  ---
29
 
@@ -34,18 +34,18 @@ Features must "fit" geometric signatures: k-simplex shapes, Cantor positions, an
34
 
35
  ## 🎯 Performance
36
 
37
- - **Best Validation Accuracy**: 79.01%
38
- - **Epoch**: 2/10
39
- - **Training Time**: 5m
40
 
41
  ### Per-Scale Performance
42
- - **Scale 576D**: 78.05%
43
- - **Scale 640D**: 77.73%
44
- - **Scale 704D**: 78.18%
45
- - **Scale 768D**: 77.52%
46
- - **Scale 832D**: 77.78%
47
- - **Scale 896D**: 77.79%
48
- - **Scale 960D**: 77.53%
49
 
50
 
51
  ## 🏗️ Architecture
@@ -73,8 +73,8 @@ Each scale learns to weight these components differently.
73
  The alpha parameter controls middle-interval weighting in the Cantor staircase.
74
 
75
  - **Initial**: 0.4746
76
- - **Final**: 0.3957
77
- - **Change**: -0.0789
78
  - **Converged to 0.5**: False
79
 
80
  The Cantor staircase uses soft triadic decomposition with learnable alpha to map
@@ -86,39 +86,39 @@ Each class has a learned scalar Cantor prototype. The model pulls features towar
86
  their class's Cantor position.
87
 
88
  **Scale 576D**:
89
- - Mean: 0.4079
90
- - Std: 0.1335
91
- - Range: [0.0001, 0.5663]
92
 
93
  **Scale 640D**:
94
- - Mean: 0.4078
95
- - Std: 0.1341
96
- - Range: [-0.0002, 0.5663]
97
 
98
  **Scale 704D**:
99
- - Mean: 0.4076
100
- - Std: 0.1266
101
- - Range: [0.0002, 0.5618]
102
 
103
  **Scale 768D**:
104
- - Mean: 0.4047
105
- - Std: 0.1384
106
- - Range: [-0.0000, 0.5645]
107
 
108
  **Scale 832D**:
109
- - Mean: 0.4062
110
- - Std: 0.1359
111
- - Range: [-0.0002, 0.5651]
112
 
113
  **Scale 896D**:
114
- - Mean: 0.4052
115
- - Std: 0.1302
116
- - Range: [-0.0001, 0.5604]
117
 
118
  **Scale 960D**:
119
- - Mean: 0.4056
120
- - Std: 0.1329
121
- - Range: [-0.0002, 0.5615]
122
 
123
 
124
  Most classes cluster around 0.5 (middle Cantor region), with smooth spread across [0,1].
@@ -128,13 +128,13 @@ This creates a continuous manifold rather than discrete bins.
128
 
129
  Each scale learns optimal weights for combining geometric components:
130
 
131
- **Scale 576D**: Feature=0.642, Cantor=0.058, Crystal=0.301
132
- **Scale 640D**: Feature=0.621, Cantor=0.058, Crystal=0.321
133
- **Scale 704D**: Feature=0.836, Cantor=0.027, Crystal=0.137
134
- **Scale 768D**: Feature=0.545, Cantor=0.071, Crystal=0.383
135
- **Scale 832D**: Feature=0.606, Cantor=0.057, Crystal=0.337
136
- **Scale 896D**: Feature=0.808, Cantor=0.028, Crystal=0.163
137
- **Scale 960D**: Feature=0.713, Cantor=0.039, Crystal=0.248
138
 
139
 
140
  **Pattern**: Lower scales rely on feature similarity, higher scales use crystal geometry.
 
23
  type: imagenet-1k
24
  metrics:
25
  - type: accuracy
26
+ value: 80.36
27
  name: Validation Accuracy
28
  ---
29
 
 
34
 
35
  ## 🎯 Performance
36
 
37
+ - **Best Validation Accuracy**: 80.36%
38
+ - **Epoch**: 4/10
39
+ - **Training Time**: 11m
40
 
41
  ### Per-Scale Performance
42
+ - **Scale 576D**: 76.15%
43
+ - **Scale 640D**: 75.49%
44
+ - **Scale 704D**: 77.49%
45
+ - **Scale 768D**: 74.66%
46
+ - **Scale 832D**: 75.15%
47
+ - **Scale 896D**: 76.66%
48
+ - **Scale 960D**: 71.48%
49
 
50
 
51
  ## 🏗️ Architecture
 
73
  The alpha parameter controls middle-interval weighting in the Cantor staircase.
74
 
75
  - **Initial**: 0.4746
76
+ - **Final**: 0.2590
77
+ - **Change**: -0.2156
78
  - **Converged to 0.5**: False
79
 
80
  The Cantor staircase uses soft triadic decomposition with learnable alpha to map
 
86
  their class's Cantor position.
87
 
88
  **Scale 576D**:
89
+ - Mean: 0.2858
90
+ - Std: 0.1704
91
+ - Range: [-0.0008, 0.5007]
92
 
93
  **Scale 640D**:
94
+ - Mean: 0.2854
95
+ - Std: 0.1706
96
+ - Range: [-0.0008, 0.5012]
97
 
98
  **Scale 704D**:
99
+ - Mean: 0.2919
100
+ - Std: 0.1680
101
+ - Range: [-0.0009, 0.5009]
102
 
103
  **Scale 768D**:
104
+ - Mean: 0.2816
105
+ - Std: 0.1721
106
+ - Range: [-0.0012, 0.5009]
107
 
108
  **Scale 832D**:
109
+ - Mean: 0.2837
110
+ - Std: 0.1713
111
+ - Range: [-0.0013, 0.5009]
112
 
113
  **Scale 896D**:
114
+ - Mean: 0.2890
115
+ - Std: 0.1691
116
+ - Range: [-0.0008, 0.5007]
117
 
118
  **Scale 960D**:
119
+ - Mean: 0.2867
120
+ - Std: 0.1701
121
+ - Range: [-0.0011, 0.5006]
122
 
123
 
124
  Most classes cluster around 0.5 (middle Cantor region), with smooth spread across [0,1].
 
128
 
129
  Each scale learns optimal weights for combining geometric components:
130
 
131
+ **Scale 576D**: Feature=0.664, Cantor=0.037, Crystal=0.300
132
+ **Scale 640D**: Feature=0.629, Cantor=0.038, Crystal=0.333
133
+ **Scale 704D**: Feature=0.936, Cantor=0.006, Crystal=0.058
134
+ **Scale 768D**: Feature=0.534, Cantor=0.047, Crystal=0.419
135
+ **Scale 832D**: Feature=0.609, Cantor=0.037, Crystal=0.354
136
+ **Scale 896D**: Feature=0.892, Cantor=0.006, Crystal=0.102
137
+ **Scale 960D**: Feature=0.737, Cantor=0.019, Crystal=0.244
138
 
139
 
140
  **Pattern**: Lower scales rely on feature similarity, higher scales use crystal geometry.
weights/GeoFractalDavid-Basin-k50/20251015_222757/model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:333a621dc2adeb89319dc56d2767c21253ba6a78ce02a512fad75a6051d23f5c
3
  size 1163870188
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a06afd2edf5ceec6759fa0cf2474a79d36e150010506952ff37522860eb25537
3
  size 1163870188
weights/GeoFractalDavid-Basin-k50/20251015_222757/model_metadata.json CHANGED
@@ -1,21 +1,21 @@
1
  {
2
- "epoch": 1,
3
  "metrics": {
4
- "val_acc": 79.014,
5
- "train_acc": 80.07714841234593,
6
  "scale_accuracies": {
7
- "576": 78.054,
8
- "640": 77.732,
9
- "704": 78.18,
10
- "768": 77.518,
11
- "832": 77.784,
12
- "896": 77.788,
13
- "960": 77.53
14
  },
15
- "best_val_acc": 79.014,
16
- "best_epoch": 1,
17
- "final_train_acc": 80.07714841234593,
18
- "training_time": "5m"
19
  },
20
  "config": {
21
  "name": "geofractal_david_basin",
@@ -72,126 +72,136 @@
72
  "alpha_summary": {
73
  "global": {
74
  "initial": 0.4745626449584961,
75
- "final": 0.39567649364471436,
76
- "change": -0.07888615131378174,
77
  "converged_to_0.5": false
78
  }
79
  },
80
  "cantor_prototypes": {
81
  "576": {
82
- "final_mean": 0.4078630805015564,
83
- "final_std": 0.1335078924894333,
84
  "final_range": [
85
- 7.831587572582066e-05,
86
- 0.5662908554077148
87
  ]
88
  },
89
  "640": {
90
- "final_mean": 0.4077507257461548,
91
- "final_std": 0.1340765506029129,
92
  "final_range": [
93
- -0.00023315568978432566,
94
- 0.5663228631019592
95
  ]
96
  },
97
  "704": {
98
- "final_mean": 0.40756258368492126,
99
- "final_std": 0.1265934258699417,
100
  "final_range": [
101
- 0.0001948980352608487,
102
- 0.5617710947990417
103
  ]
104
  },
105
  "768": {
106
- "final_mean": 0.40472421050071716,
107
- "final_std": 0.13841961324214935,
108
  "final_range": [
109
- -1.0782070603454486e-05,
110
- 0.5644935965538025
111
  ]
112
  },
113
  "832": {
114
- "final_mean": 0.40620630979537964,
115
- "final_std": 0.1359475702047348,
116
  "final_range": [
117
- -0.0002186586061725393,
118
- 0.5651434063911438
119
  ]
120
  },
121
  "896": {
122
- "final_mean": 0.40519979596138,
123
- "final_std": 0.1301906853914261,
124
  "final_range": [
125
- -0.00011574607196962461,
126
- 0.5603952407836914
127
  ]
128
  },
129
  "960": {
130
- "final_mean": 0.4056411385536194,
131
- "final_std": 0.1329261064529419,
132
  "final_range": [
133
- -0.00016488014080096036,
134
- 0.5614519715309143
135
  ]
136
  }
137
  },
138
  "geo_weights": {
139
  "576": {
140
- "feature": 0.6415244936943054,
141
- "cantor": 0.05753276124596596,
142
- "crystal": 0.3009428381919861
143
  },
144
  "640": {
145
- "feature": 0.621259868144989,
146
- "cantor": 0.057620953768491745,
147
- "crystal": 0.3211192190647125
148
  },
149
  "704": {
150
- "feature": 0.8358007669448853,
151
- "cantor": 0.027461165562272072,
152
- "crystal": 0.13673809170722961
153
  },
154
  "768": {
155
- "feature": 0.5451870560646057,
156
- "cantor": 0.07136273384094238,
157
- "crystal": 0.3834501802921295
158
  },
159
  "832": {
160
- "feature": 0.6060996055603027,
161
- "cantor": 0.057310495525598526,
162
- "crystal": 0.3365899324417114
163
  },
164
  "896": {
165
- "feature": 0.8084424138069153,
166
- "cantor": 0.028363272547721863,
167
- "crystal": 0.1631942242383957
168
  },
169
  "960": {
170
- "feature": 0.7132336497306824,
171
- "cantor": 0.03874538466334343,
172
- "crystal": 0.2480209320783615
173
  }
174
  },
175
  "training_history": {
176
  "epochs": [
177
  1,
178
- 2
 
 
179
  ],
180
  "train_loss": [
181
  1.9556559649900125,
182
- 1.4780290030633298
 
 
183
  ],
184
  "train_acc": [
185
  73.77352054806282,
186
- 80.07714841234593
 
 
187
  ],
188
  "val_acc": [
189
  77.872,
190
- 79.014
 
 
191
  ],
192
  "lr": [
193
  0.001,
194
- 0.0009755527298894294
 
 
195
  ]
196
  }
197
  }
 
1
  {
2
+ "epoch": 3,
3
  "metrics": {
4
+ "val_acc": 80.356,
5
+ "train_acc": 81.860054153752,
6
  "scale_accuracies": {
7
+ "576": 76.152,
8
+ "640": 75.488,
9
+ "704": 77.492,
10
+ "768": 74.662,
11
+ "832": 75.152,
12
+ "896": 76.66,
13
+ "960": 71.48
14
  },
15
+ "best_val_acc": 80.356,
16
+ "best_epoch": 3,
17
+ "final_train_acc": 81.860054153752,
18
+ "training_time": "11m"
19
  },
20
  "config": {
21
  "name": "geofractal_david_basin",
 
72
  "alpha_summary": {
73
  "global": {
74
  "initial": 0.4745626449584961,
75
+ "final": 0.25898104906082153,
76
+ "change": -0.21558159589767456,
77
  "converged_to_0.5": false
78
  }
79
  },
80
  "cantor_prototypes": {
81
  "576": {
82
+ "final_mean": 0.28581517934799194,
83
+ "final_std": 0.17039097845554352,
84
  "final_range": [
85
+ -0.0007920649368315935,
86
+ 0.5007304549217224
87
  ]
88
  },
89
  "640": {
90
+ "final_mean": 0.28536349534988403,
91
+ "final_std": 0.1706337034702301,
92
  "final_range": [
93
+ -0.0008051643380895257,
94
+ 0.5011658072471619
95
  ]
96
  },
97
  "704": {
98
+ "final_mean": 0.2919076979160309,
99
+ "final_std": 0.16795963048934937,
100
  "final_range": [
101
+ -0.0008631697855889797,
102
+ 0.5009223818778992
103
  ]
104
  },
105
  "768": {
106
+ "final_mean": 0.2815600633621216,
107
+ "final_std": 0.17207278311252594,
108
  "final_range": [
109
+ -0.0012111648684367537,
110
+ 0.5009366273880005
111
  ]
112
  },
113
  "832": {
114
+ "final_mean": 0.28371378779411316,
115
+ "final_std": 0.17125703394412994,
116
  "final_range": [
117
+ -0.0012689391151070595,
118
+ 0.5008506178855896
119
  ]
120
  },
121
  "896": {
122
+ "final_mean": 0.2890189290046692,
123
+ "final_std": 0.16906769573688507,
124
  "final_range": [
125
+ -0.0007662388379685581,
126
+ 0.5006508231163025
127
  ]
128
  },
129
  "960": {
130
+ "final_mean": 0.2866990566253662,
131
+ "final_std": 0.17010973393917084,
132
  "final_range": [
133
+ -0.0010537795023992658,
134
+ 0.5005902647972107
135
  ]
136
  }
137
  },
138
  "geo_weights": {
139
  "576": {
140
+ "feature": 0.6637648344039917,
141
+ "cantor": 0.03667046129703522,
142
+ "crystal": 0.29956477880477905
143
  },
144
  "640": {
145
+ "feature": 0.6294592618942261,
146
+ "cantor": 0.037632085382938385,
147
+ "crystal": 0.3329086899757385
148
  },
149
  "704": {
150
+ "feature": 0.9357029795646667,
151
+ "cantor": 0.005980553571134806,
152
+ "crystal": 0.058316417038440704
153
  },
154
  "768": {
155
+ "feature": 0.5343316197395325,
156
+ "cantor": 0.04711243882775307,
157
+ "crystal": 0.41855594515800476
158
  },
159
  "832": {
160
+ "feature": 0.6087399125099182,
161
+ "cantor": 0.037428487092256546,
162
+ "crystal": 0.35383158922195435
163
  },
164
  "896": {
165
+ "feature": 0.8923861384391785,
166
+ "cantor": 0.0058705611154437065,
167
+ "crystal": 0.10174331068992615
168
  },
169
  "960": {
170
+ "feature": 0.736908495426178,
171
+ "cantor": 0.018661215901374817,
172
+ "crystal": 0.24443036317825317
173
  }
174
  },
175
  "training_history": {
176
  "epochs": [
177
  1,
178
+ 2,
179
+ 3,
180
+ 4
181
  ],
182
  "train_loss": [
183
  1.9556559649900125,
184
+ 1.4780290030633298,
185
+ 1.3460614071867336,
186
+ 1.2537885870034702
187
  ],
188
  "train_acc": [
189
  73.77352054806282,
190
+ 80.07714841234593,
191
+ 81.04048886679098,
192
+ 81.860054153752
193
  ],
194
  "val_acc": [
195
  77.872,
196
+ 79.014,
197
+ 79.78,
198
+ 80.356
199
  ],
200
  "lr": [
201
  0.001,
202
+ 0.0009755527298894294,
203
+ 0.0009046039886902864,
204
+ 0.0007940987335200905
205
  ]
206
  }
207
  }
weights/GeoFractalDavid-Basin-k50/20251015_222757/training_history.json CHANGED
@@ -2,91 +2,133 @@
2
  "training_history": {
3
  "epochs": [
4
  1,
5
- 2
 
 
6
  ],
7
  "train_loss": [
8
  1.9556559649900125,
9
- 1.4780290030633298
 
 
10
  ],
11
  "train_acc": [
12
  73.77352054806282,
13
- 80.07714841234593
 
 
14
  ],
15
  "val_acc": [
16
  77.872,
17
- 79.014
 
 
18
  ],
19
  "lr": [
20
  0.001,
21
- 0.0009755527298894294
 
 
22
  ]
23
  },
24
  "loss_components": {
25
  "contrastive": [
26
  1.842834744685755,
27
- 1.3882114456865353
 
 
28
  ],
29
  "correct": [
30
  0.6731035322807848,
31
- 0.5506705783140926
 
 
32
  ],
33
  "incorrect": [
34
  0.44108464837836,
35
- 0.4929031326224248
 
 
36
  ],
37
  "contrast": [
38
  1.898377784715293,
39
- 1.1821785980520156
 
 
40
  ],
41
  "coherence": [
42
  0.1728733149631308,
43
- 0.13558272026574436
 
 
44
  ],
45
  "separation": [
46
  0.01890343071111583,
47
- 0.033916719134136984
 
 
48
  ],
49
  "discretization": [
50
  0.10356761047991515,
51
- 0.059255850908998195
 
 
52
  ],
53
  "total": [
54
  1.9556559649900125,
55
- 1.4780290030633298
 
 
56
  ]
57
  },
58
  "scale_accuracies": {
59
  "576": [
60
  77.334,
61
- 78.054
 
 
62
  ],
63
  "640": [
64
  77.12,
65
- 77.732
 
 
66
  ],
67
  "704": [
68
  77.138,
69
- 78.18
 
 
70
  ],
71
  "768": [
72
  77.28,
73
- 77.518
 
 
74
  ],
75
  "832": [
76
  77.316,
77
- 77.784
 
 
78
  ],
79
  "896": [
80
  77.114,
81
- 77.788
 
 
82
  ],
83
  "960": [
84
  77.104,
85
- 77.53
 
 
86
  ]
87
  },
88
  "alpha_history": [
89
  0.4745626449584961,
90
- 0.39567649364471436
 
 
91
  ]
92
  }
 
2
  "training_history": {
3
  "epochs": [
4
  1,
5
+ 2,
6
+ 3,
7
+ 4
8
  ],
9
  "train_loss": [
10
  1.9556559649900125,
11
+ 1.4780290030633298,
12
+ 1.3460614071867336,
13
+ 1.2537885870034702
14
  ],
15
  "train_acc": [
16
  73.77352054806282,
17
+ 80.07714841234593,
18
+ 81.04048886679098,
19
+ 81.860054153752
20
  ],
21
  "val_acc": [
22
  77.872,
23
+ 79.014,
24
+ 79.78,
25
+ 80.356
26
  ],
27
  "lr": [
28
  0.001,
29
+ 0.0009755527298894294,
30
+ 0.0009046039886902864,
31
+ 0.0007940987335200905
32
  ]
33
  },
34
  "loss_components": {
35
  "contrastive": [
36
  1.842834744685755,
37
+ 1.3882114456865353,
38
+ 1.2652259323353203,
39
+ 1.1790547934583961
40
  ],
41
  "correct": [
42
  0.6731035322807848,
43
+ 0.5506705783140926,
44
+ 0.4550487843517678,
45
+ 0.4209542488232969
46
  ],
47
  "incorrect": [
48
  0.44108464837836,
49
+ 0.4929031326224248,
50
+ 0.5791181146908111,
51
+ 0.6085665392132994
52
  ],
53
  "contrast": [
54
  1.898377784715293,
55
+ 1.1821785980520156,
56
+ 1.0412361860370483,
57
+ 0.9076345494856088
58
  ],
59
  "coherence": [
60
  0.1728733149631308,
61
+ 0.13558272026574436,
62
+ 0.11571173029235532,
63
+ 0.09865620135597866
64
  ],
65
  "separation": [
66
  0.01890343071111583,
67
+ 0.033916719134136984,
68
+ 0.030128026784736462,
69
+ 0.023460495624893607
70
  ],
71
  "discretization": [
72
  0.10356761047991515,
73
+ 0.059255850908998195,
74
+ 0.06970599833799387,
75
+ 0.09183770244314568
76
  ],
77
  "total": [
78
  1.9556559649900125,
79
+ 1.4780290030633298,
80
+ 1.3460614071867336,
81
+ 1.2537885870034702
82
  ]
83
  },
84
  "scale_accuracies": {
85
  "576": [
86
  77.334,
87
+ 78.054,
88
+ 77.45,
89
+ 76.152
90
  ],
91
  "640": [
92
  77.12,
93
+ 77.732,
94
+ 76.824,
95
+ 75.488
96
  ],
97
  "704": [
98
  77.138,
99
+ 78.18,
100
+ 78.17,
101
+ 77.492
102
  ],
103
  "768": [
104
  77.28,
105
+ 77.518,
106
+ 76.376,
107
+ 74.662
108
  ],
109
  "832": [
110
  77.316,
111
+ 77.784,
112
+ 76.732,
113
+ 75.152
114
  ],
115
  "896": [
116
  77.114,
117
+ 77.788,
118
+ 76.992,
119
+ 76.66
120
  ],
121
  "960": [
122
  77.104,
123
+ 77.53,
124
+ 74.826,
125
+ 71.48
126
  ]
127
  },
128
  "alpha_history": [
129
  0.4745626449584961,
130
+ 0.39567649364471436,
131
+ 0.33655133843421936,
132
+ 0.25898104906082153
133
  ]
134
  }