Update README - GeoFractalDavid-Basin-k7 - Run 20251016_001155 - Acc 67.72%

README.md

@@ -14,7 +14,7 @@ metrics:
 - accuracy
 library_name: pytorch
 model-index:
-- name: GeoFractalDavid-Basin-k9
+- name: GeoFractalDavid-Basin-k7
   results:
   - task:
       type: image-classification
@@ -23,31 +23,29 @@ model-index:
       type: imagenet-1k
     metrics:
     - type: accuracy
-      value: 78.89
+      value: 67.72
       name: Validation Accuracy
 ---
 
-# GeoFractalDavid-Basin-k9: Geometric Basin Classification
+# GeoFractalDavid-Basin-k7: Geometric Basin Classification
 
 **GeoFractalDavid** achieves classification through geometric compatibility rather than cross-entropy.
 Features must "fit" geometric signatures: k-simplex shapes, Cantor positions, and hierarchical structure.
 
 ## 🎯 Performance
 
-- **Best Validation Accuracy**: 78.89%
+- **Best Validation Accuracy**: 67.72%
 - **Epoch**: 2/10
-- **Training Time**: 5m
+- **Training Time**: 3m
 
 ### Per-Scale Performance
-- **Scale 512D**: 78.11%
-- **Scale 576D**: 78.22%
-- **Scale 640D**: 78.29%
-- **Scale 704D**: 78.22%
-- **Scale 768D**: 78.10%
-- **Scale 832D**: 78.11%
-- **Scale 896D**: 77.50%
-- **Scale 960D**: 77.48%
-- **Scale 1024D**: 76.10%
+- **Scale 320D**: 66.08%
+- **Scale 384D**: 66.29%
+- **Scale 448D**: 66.44%
+- **Scale 512D**: 66.57%
+- **Scale 576D**: 66.48%
+- **Scale 640D**: 65.66%
+- **Scale 704D**: 64.23%
 
 
 ## 🏗️ Architecture
@@ -55,11 +53,11 @@ Features must "fit" geometric signatures: k-simplex shapes, Cantor positions, an
 **Model Type**: Multi-scale geometric basin classifier
 
 **Core Components**:
-- **Feature Dimension**: 768
+- **Feature Dimension**: 512
 - **Number of Classes**: 1000
-- **k-Simplex Structure**: k=9 (10 vertices per class)
-- **Scales**: [512, 576, 640, 704, 768, 832, 896, 960, 1024]
-- **Total Simplex Vertices**: 10,000
+- **k-Simplex Structure**: k=7 (8 vertices per class)
+- **Scales**: [320, 384, 448, 512, 576, 640, 704]
+- **Total Simplex Vertices**: 8,000
 
 **Geometric Components**:
 1. **Feature Similarity**: Cosine similarity to k-simplex centroids
@@ -74,10 +72,10 @@ Each scale learns to weight these components differently.
 
 The alpha parameter controls middle-interval weighting in the Cantor staircase.
 
-- **Initial**: 0.5078
-- **Final**: 0.5387
-- **Change**: +0.0309
-- **Converged to 0.5**: True
+- **Initial**: 0.1869
+- **Final**: 0.2275
+- **Change**: +0.0406
+- **Converged to 0.5**: False
 
 The Cantor staircase uses soft triadic decomposition with learnable alpha to map
 features into [0,1] space with fractal structure.
@@ -87,50 +85,40 @@ features into [0,1] space with fractal structure.
 Each class has a learned scalar Cantor prototype. The model pulls features toward
 their class's Cantor position.
 
+**Scale 320D**:
+- Mean: 0.2813
+- Std: 0.1235
+- Range: [0.0390, 0.4998]
+
+**Scale 384D**:
+- Mean: 0.2804
+- Std: 0.1237
+- Range: [0.0377, 0.4998]
+
+**Scale 448D**:
+- Mean: 0.2801
+- Std: 0.1237
+- Range: [0.0368, 0.4998]
+
 **Scale 512D**:
-- Mean: 0.5389
-- Std: 0.1279
-- Range: [0.2747, 0.7762]
+- Mean: 0.2969
+- Std: 0.1208
+- Range: [0.0492, 0.5041]
 
 **Scale 576D**:
-- Mean: 0.5391
-- Std: 0.1278
-- Range: [0.2753, 0.7758]
+- Mean: 0.2965
+- Std: 0.1208
+- Range: [0.0501, 0.5020]
 
 **Scale 640D**:
-- Mean: 0.5391
-- Std: 0.1277
-- Range: [0.2756, 0.7758]
+- Mean: 0.2906
+- Std: 0.1219
+- Range: [0.0455, 0.5000]
 
 **Scale 704D**:
-- Mean: 0.5384
-- Std: 0.1274
-- Range: [0.2757, 0.7740]
-
-**Scale 768D**:
-- Mean: 0.5344
-- Std: 0.1296
-- Range: [0.2642, 0.7727]
-
-**Scale 832D**:
-- Mean: 0.5376
-- Std: 0.1279
-- Range: [0.2729, 0.7738]
-
-**Scale 896D**:
-- Mean: 0.5361
-- Std: 0.1295
-- Range: [0.2662, 0.7758]
-
-**Scale 960D**:
-- Mean: 0.5367
-- Std: 0.1287
-- Range: [0.2695, 0.7749]
-
-**Scale 1024D**:
-- Mean: 0.5375
-- Std: 0.1283
-- Range: [0.2718, 0.7747]
+- Mean: 0.2902
+- Std: 0.1219
+- Range: [0.0453, 0.4998]
 
 
 Most classes cluster around 0.5 (middle Cantor region), with smooth spread across [0,1].
@@ -140,15 +128,13 @@ This creates a continuous manifold rather than discrete bins.
 
 Each scale learns optimal weights for combining geometric components:
 
-**Scale 512D**: Feature=0.692, Cantor=0.072, Crystal=0.235
-**Scale 576D**: Feature=0.651, Cantor=0.073, Crystal=0.276
-**Scale 640D**: Feature=0.619, Cantor=0.074, Crystal=0.307
-**Scale 704D**: Feature=0.626, Cantor=0.066, Crystal=0.308
-**Scale 768D**: Feature=0.829, Cantor=0.031, Crystal=0.140
-**Scale 832D**: Feature=0.694, Cantor=0.048, Crystal=0.258
-**Scale 896D**: Feature=0.802, Cantor=0.032, Crystal=0.166
-**Scale 960D**: Feature=0.735, Cantor=0.038, Crystal=0.227
-**Scale 1024D**: Feature=0.663, Cantor=0.042, Crystal=0.295
+**Scale 320D**: Feature=0.732, Cantor=0.085, Crystal=0.183
+**Scale 384D**: Feature=0.713, Cantor=0.087, Crystal=0.200
+**Scale 448D**: Feature=0.675, Cantor=0.091, Crystal=0.234
+**Scale 512D**: Feature=0.863, Cantor=0.033, Crystal=0.104
+**Scale 576D**: Feature=0.812, Cantor=0.036, Crystal=0.152
+**Scale 640D**: Feature=0.749, Cantor=0.042, Crystal=0.210
+**Scale 704D**: Feature=0.701, Cantor=0.041, Crystal=0.258
 
 
 **Pattern**: Lower scales rely on feature similarity, higher scales use crystal geometry.