Epoch 50: 55.79%

README.md

@@ -5,7 +5,7 @@ tags:
 - cifar100
 - geometric-learning
 - fractal-encoding
-- trained
+- in-training
 - no-attention
 - no-cross-entropy
 datasets:
@@ -15,7 +15,7 @@ metrics:
 library_name: pytorch
 pipeline_tag: image-classification
 model-index:
-- name: geo-beatrix-step4-feats100
+- name: geo-beatrix-fractal
   results:
   - task:
       type: image-classification
@@ -30,13 +30,13 @@ model-index:
       verified: false
 ---
 
-# geo-beatrix-step4-feats100
+# geo-beatrix-fractal
 
 **Geometric Basin Classification for CIFAR-100**
 
-🎉 **Training Complete** 🎉
+🚧 **Training in Progress** 🚧
 
-Final Status: Epoch 200/200
+Current Status: Epoch 50/200
 
 ---
 
@@ -46,26 +46,9 @@ Final Status: Epoch 200/200
 |--------|-------|
 | **Best Test Accuracy** | **69.08%** |
 | **Best Epoch** | 190 |
-| **Current Train Accuracy** | 68.80% |
-| **Current Test Accuracy** | 60.81% |
-| **Current α (Cantor param)** | 0.4320 |
-| **Total Parameters** | 45,356,337 |
-| **Training Time** | 0:48:43 |
-
-### All Training Runs
-
-| Timestamp | Status | Best Epoch | Test Acc | Train Acc | α |
-|-----------|--------|------------|----------|-----------|---|
-| `20251010_154433` | ✅ | 190 | **69.08%** | 68.80% | 0.4320 |
-
-### Comparison to State-of-the-Art
-
-| Model | Accuracy | Status |
-|-------|----------|--------|
-| **geo-beatrix (this model)** | **69.08%** | ✅ Complete |
-| vit-beatrix-dualstream | 66.0% | Vision Transformer + Cross-Entropy |
-
-✅ **geo-beatrix has surpassed all baselines!**
+| **Current α (Cantor param)** | 0.4165 |
+| **Total Parameters** | 45,161,489 |
+| **Mixing Mode** | Fractal (triadic) |
 
 ---
 
@@ -74,137 +57,11 @@ 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**: 4
-- **PE Features/Level**: 100
-- **Classification**: Geometric Basin Compatibility (NO cross-entropy)
-- **Attention Mechanisms**: NONE
-
----
-
-## Training Configuration
-
-```json
-{
-  "model_name": "geo-beatrix-step4-feats100",
-  "model_type": "geometric_basin_classifier",
-  "num_classes": 100,
-  "batch_size": 256,
-  "num_epochs": 200,
-  "base_learning_rate": 0.001,
-  "weight_decay": 0.05,
-  "warmup_epochs": 10,
-  "pe_levels": 4,
-  "pe_features_per_level": 100,
-  "dropout": 0.1,
-  "upload_every_n_epochs": 50,
-  "alphamix": {
-    "enabled": true,
-    "range": [
-      0.3,
-      0.7
-    ],
-    "spatial_ratio": 0.25,
-    "curriculum_start": 0.05,
-    "curriculum_end": 0.25
-  },
-  "architecture": "ResNet-style with Devil's Staircase PE",
-  "loss_function": "Geometric Basin Compatibility",
-  "cross_entropy": false,
-  "attention_mechanisms": false,
-  "timestamp": "20251010_154433"
-}
-```
-
----
-
-## Files Structure
-
-```
-├── model.pt                 (BEST overall model - easy access!)
-├── model.safetensors        (BEST overall model - easy access!)
-├── best_model_info.json     (which epoch/run this came from)
-├── runs_history.json        (all training runs and their results)
-├── README.md
-├── weights/geo-beatrix-step4-feats100/20251010_154433/
-│   ├── model.pt                 (best from this training run)
-│   ├── model.safetensors        (best from this training run)
-│   ├── config.json
-│   ├── training_log.txt
-│   └── checkpoints/
-│       ├── checkpoint_epoch_10.safetensors
-│       ├── checkpoint_epoch_20.safetensors
-│       └── checkpoint_epoch_30.safetensors
-│       (snapshots every 50 epochs)
-└── runs/geo-beatrix-step4-feats100/20251010_154433/
-    ├── events.out.tfevents.*    (TensorBoard logs)
-    └── metrics.csv              (training metrics)
-```
-
-**Note**: The root `model.pt` and `model.safetensors` always contain the best model across all training runs!
-
----
-
-## Usage
-
-```python
-from huggingface_hub import hf_hub_download
-import torch
-
-# EASIEST: Download BEST overall model from root (recommended!)
-from safetensors.torch import load_file
-model_path = hf_hub_download(
-    repo_id="AbstractPhil/geo-beatrix",
-    filename="model.safetensors"
-)
-state_dict = load_file(model_path)
-# model.load_state_dict(state_dict)
-
-# Check which epoch/run the best model came from
-info_path = hf_hub_download(
-    repo_id="AbstractPhil/geo-beatrix",
-    filename="best_model_info.json"
-)
-with open(info_path) as f:
-    best_info = json.load(f)
-    print(f"Best model: epoch {best_info['epoch']}, {best_info['test_accuracy']:.2f}%")
-
-# Or download from specific training run
-model_path = hf_hub_download(
-    repo_id="AbstractPhil/geo-beatrix",
-    filename="weights/geo-beatrix-step4-feats100/20251010_154433/model.safetensors"
-)
-
-# Download specific epoch checkpoint
-epoch_checkpoint = hf_hub_download(
-    repo_id="AbstractPhil/geo-beatrix",
-    filename="weights/geo-beatrix-step4-feats100/20251010_154433/checkpoints/checkpoint_epoch_100.safetensors"
-)
-```
-
----
-
-## Training History
-
-### Best Checkpoint
-- Epoch: 190
-- Train Acc: 64.35%
-- Test Acc: 60.00%
-- Alpha: 0.4318
-- Loss: 0.9990
-
-### Latest 5 Epochs
-
-- **Epoch 196**: Train 64.06%, Test 0.00%, α=0.4320, Loss=0.9867
-- **Epoch 197**: Train 67.90%, Test 0.00%, α=0.4320, Loss=1.0384
-- **Epoch 198**: Train 64.27%, Test 0.00%, α=0.4320, Loss=0.9999
-- **Epoch 199**: Train 65.64%, Test 0.00%, α=0.4320, Loss=1.0067
-- **Epoch 200**: Train 68.80%, Test 60.81%, α=0.4320, Loss=1.0461
-
-### Training Milestones
-- 🎯 **50% Accuracy** reached at epoch 35
-- 🎯 **60% Accuracy** reached at epoch 135
-- 📊 **α ≥ 0.40** reached at epoch 8
-- 📊 **α ≥ 0.44** (near triadic equilibrium) at epoch 11
+- **PE Levels**: 20
+- **PE Features/Level**: 4
+- **Classification**: Geometric Basin Compatibility
+- **Attention**: NONE
+- **Cross-Entropy**: NONE
 
 ---
 
@@ -214,7 +71,7 @@ epoch_checkpoint = hf_hub_download(
 ✅ **NO cross-entropy loss**  
 ✅ **Fractal positional encoding** (Cantor function from 1883)  
 ✅ **Geometric compatibility classification**  
-✅ **Ancient convolutions** (ResNet-style from 1990s)
+✅ **Triadic fractal mixing** (base-3 aligned)
 
 ---