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 (legacy)
Model card Files
xet
Metrics Community
AbstractPhil commited on
Commit
563494b
·
verified ·
1 Parent(s): daf261f

Epoch 200: 64.32%

Browse files
Files changed (1) hide show
  1. README.md +13 -13
README.md CHANGED
@@ -5,7 +5,7 @@ tags:
5
  - cifar100
6
  - geometric-learning
7
  - fractal-encoding
8
- - in-training
9
  - no-attention
10
  - no-cross-entropy
11
  datasets:
@@ -34,9 +34,9 @@ model-index:
34
 
35
  **Geometric Basin Classification for CIFAR-100**
36
 
37
- 🚧 **Training in Progress** 🚧
38
 
39
- Current Status: Epoch 190/200
40
 
41
  ---
42
 
@@ -46,17 +46,17 @@ Current Status: Epoch 190/200
46
  |--------|-------|
47
  | **Best Test Accuracy** | **67.00%** |
48
  | **Best Epoch** | 190 |
49
- | **Current Train Accuracy** | 61.47% |
50
- | **Current Test Accuracy** | 67.00% |
51
- | **Current α (Cantor param)** | 0.4425 |
52
  | **Total Parameters** | 45,161,489 |
53
- | **Training Time** | 1:06:19 |
54
 
55
  ### Comparison to State-of-the-Art
56
 
57
  | Model | Accuracy | Status |
58
  |-------|----------|--------|
59
- | **geo-beatrix (this model)** | **67.00%** | 🔄 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 |
@@ -184,11 +184,11 @@ tensorboard_files = hf_hub_download(
184
 
185
  ### Latest 5 Epochs
186
 
187
- - **Epoch 186**: Train 65.21%, Test 0.00%, α=0.4426, Loss=0.5107
188
- - **Epoch 187**: Train 66.32%, Test 0.00%, α=0.4425, Loss=0.5147
189
- - **Epoch 188**: Train 63.06%, Test 0.00%, α=0.4425, Loss=0.4994
190
- - **Epoch 189**: Train 62.33%, Test 0.00%, α=0.4425, Loss=0.5007
191
- - **Epoch 190**: Train 61.47%, Test 67.00%, α=0.4425, Loss=0.5031
192
 
193
  ### Training Milestones
194
  - 🎯 **50% Accuracy** reached at epoch 45
 
5
  - cifar100
6
  - geometric-learning
7
  - fractal-encoding
8
+ - trained
9
  - no-attention
10
  - no-cross-entropy
11
  datasets:
 
34
 
35
  **Geometric Basin Classification for CIFAR-100**
36
 
37
+ 🎉 **Training Complete** 🎉
38
 
39
+ Final Status: Epoch 200/200
40
 
41
  ---
42
 
 
46
  |--------|-------|
47
  | **Best Test Accuracy** | **67.00%** |
48
  | **Best Epoch** | 190 |
49
+ | **Current Train Accuracy** | 63.49% |
50
+ | **Current Test Accuracy** | 64.32% |
51
+ | **Current α (Cantor param)** | 0.4422 |
52
  | **Total Parameters** | 45,161,489 |
53
+ | **Training Time** | 1:09:46 |
54
 
55
  ### Comparison to State-of-the-Art
56
 
57
  | Model | Accuracy | Status |
58
  |-------|----------|--------|
59
+ | **geo-beatrix (this model)** | **67.00%** | ✅ Complete |
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 |
 
184
 
185
  ### Latest 5 Epochs
186
 
187
+ - **Epoch 196**: Train 63.32%, Test 0.00%, α=0.4422, Loss=0.4936
188
+ - **Epoch 197**: Train 65.20%, Test 0.00%, α=0.4422, Loss=0.5069
189
+ - **Epoch 198**: Train 60.80%, Test 0.00%, α=0.4422, Loss=0.4929
190
+ - **Epoch 199**: Train 58.87%, Test 0.00%, α=0.4422, Loss=0.4874
191
+ - **Epoch 200**: Train 63.49%, Test 64.32%, α=0.4422, Loss=0.5049
192
 
193
  ### Training Milestones
194
  - 🎯 **50% Accuracy** reached at epoch 45