Epoch 50: 57.67%

README.md

@@ -5,7 +5,7 @@ tags:
 - cifar100
 - geometric-learning
 - fractal-encoding
-- trained
+- in-training
 - no-attention
 - no-cross-entropy
 datasets:
@@ -34,9 +34,9 @@ model-index:
 
 **Geometric Basin Classification for CIFAR-100**
 
-🎉 **Training Complete** 🎉
+🚧 **Training in Progress** 🚧
 
-Final Status: Epoch 200/200
+Current Status: Epoch 50/200
 
 ---
 
@@ -46,23 +46,23 @@ Final Status: Epoch 200/200
 |--------|-------|
 | **Best Test Accuracy** | **66.72%** |
 | **Best Epoch** | 190 |
-| **Current Train Accuracy** | 45.28% |
-| **Current Test Accuracy** | 51.26% |
-| **Current α (Cantor param)** | 0.4260 |
-| **Total Parameters** | 45,172,297 |
-| **Training Time** | 0:50:42 |
+| **Current Train Accuracy** | 61.38% |
+| **Current Test Accuracy** | 57.67% |
+| **Current α (Cantor param)** | 0.4204 |
+| **Total Parameters** | 45,235,067 |
+| **Training Time** | 0:13:10 |
 
 ### All Training Runs
 
 | Timestamp | Status | Best Epoch | Test Acc | Train Acc | α |
 |-----------|--------|------------|----------|-----------|---|
-| `20251010_004820` | ✅ | 190 | **66.72%** | 45.28% | 0.4260 |
+| `20251010_014649` | 🔄 | 190 | **66.72%** | 61.38% | 0.4204 |
 
 ### Comparison to State-of-the-Art
 
 | Model | Accuracy | Status |
 |-------|----------|--------|
-| **geo-beatrix (this model)** | **66.72%** | ✅ Complete |
+| **geo-beatrix (this model)** | **66.72%** | 🔄 Training |
 | vit-beatrix-dualstream | 66.0% | Vision Transformer + Cross-Entropy |
 | CLIP ViT-L/14 (zero-shot) | ~63-65% | 400M image-text pairs |
 | CLIP ViT-B/32 (zero-shot) | ~63.5% | Vision Transformer |
@@ -76,8 +76,8 @@ Final Status: Epoch 200/200
 - **Base**: ResNet-style with residual blocks
 - **Channels**: 64 → 128 → 256 → 512 → 1024
 - **Positional Encoding**: Devil's Staircase (Cantor function, 1883)
-- **PE Levels**: 5
-- **PE Features/Level**: 20
+- **PE Levels**: 20
+- **PE Features/Level**: 10
 - **Classification**: Geometric Basin Compatibility (NO cross-entropy)
 - **Attention Mechanisms**: NONE
 
@@ -95,8 +95,8 @@ Final Status: Epoch 200/200
   "base_learning_rate": 0.001,
   "weight_decay": 0.05,
   "warmup_epochs": 10,
-  "pe_levels": 5,
-  "pe_features_per_level": 20,
+  "pe_levels": 20,
+  "pe_features_per_level": 10,
   "dropout": 0.1,
   "upload_every_n_epochs": 50,
   "alphamix": {
@@ -113,7 +113,7 @@ Final Status: Epoch 200/200
   "loss_function": "Geometric Basin Compatibility",
   "cross_entropy": false,
   "attention_mechanisms": false,
-  "timestamp": "20251010_004820"
+  "timestamp": "20251010_014649"
 }
 ```
 
@@ -127,7 +127,7 @@ Final Status: Epoch 200/200
 ├── best_model_info.json     (which epoch/run this came from)
 ├── runs_history.json        (all training runs and their results)
 ├── README.md
-├── weights/geo-beatrix/20251010_004820/
+├── weights/geo-beatrix/20251010_014649/
 │   ├── model.pt                 (best from this training run)
 │   ├── model.safetensors        (best from this training run)
 │   ├── config.json
@@ -137,7 +137,7 @@ Final Status: Epoch 200/200
 │       ├── checkpoint_epoch_20.safetensors
 │       └── checkpoint_epoch_30.safetensors
 │       (snapshots every 50 epochs)
-└── runs/geo-beatrix/20251010_004820/
+└── runs/geo-beatrix/20251010_014649/
     ├── events.out.tfevents.*    (TensorBoard logs)
     └── metrics.csv              (training metrics)
 ```
@@ -173,13 +173,13 @@ with open(info_path) as f:
 # Or download from specific training run
 model_path = hf_hub_download(
     repo_id="AbstractPhil/geo-beatrix",
-    filename="weights/geo-beatrix/20251010_004820/model.safetensors"
+    filename="weights/geo-beatrix/20251010_014649/model.safetensors"
 )
 
 # Download specific epoch checkpoint
 epoch_checkpoint = hf_hub_download(
     repo_id="AbstractPhil/geo-beatrix",
-    filename="weights/geo-beatrix/20251010_004820/checkpoints/checkpoint_epoch_100.safetensors"
+    filename="weights/geo-beatrix/20251010_014649/checkpoints/checkpoint_epoch_100.safetensors"
 )
 ```
 
@@ -189,22 +189,21 @@ epoch_checkpoint = hf_hub_download(
 
 ### Best Checkpoint
 - Epoch: 190
-- Train Acc: 51.34%
-- Test Acc: 56.68%
-- Alpha: 0.4259
-- Loss: 0.8149
+- Train Acc: 61.38%
+- Test Acc: 66.72%
+- Alpha: 0.4204
+- Loss: 0.0000
 
 ### Latest 5 Epochs
 
-- **Epoch 196**: Train 49.67%, Test 0.00%, α=0.4260, Loss=0.8037
-- **Epoch 197**: Train 46.56%, Test 0.00%, α=0.4260, Loss=0.7564
-- **Epoch 198**: Train 50.45%, Test 0.00%, α=0.4260, Loss=0.8134
-- **Epoch 199**: Train 53.19%, Test 0.00%, α=0.4260, Loss=0.8330
-- **Epoch 200**: Train 45.28%, Test 51.26%, α=0.4260, Loss=0.7648
+- **Epoch 46**: Train 59.76%, Test 0.00%, α=0.4153, Loss=1.2259
+- **Epoch 47**: Train 59.94%, Test 0.00%, α=0.4171, Loss=1.1919
+- **Epoch 48**: Train 61.30%, Test 0.00%, α=0.4238, Loss=1.2102
+- **Epoch 49**: Train 61.85%, Test 0.00%, α=0.4170, Loss=1.1685
+- **Epoch 50**: Train 61.38%, Test 57.67%, α=0.4204, Loss=1.1308
 
 ### Training Milestones
-- 🎯 **50% Accuracy** reached at epoch 30
-- 🎯 **60% Accuracy** reached at epoch 95
+- 🎯 **50% Accuracy** reached at epoch 40
 - 📊 **α ≥ 0.40** reached at epoch 9
 
 ---