Update beatrix-cifar100_20251007_203741 - Epoch 0 - Acc: 0.0470

README.md

@@ -14,6 +14,28 @@ license: mit
 This repository contains Vision Transformers integrating Devil's Staircase positional encoding 
 with geometric simplex features for vision tasks.
 
+## Key Features
+
+- **Fractal Positional Encoding**: Devil's Staircase multi-scale position embeddings
+- **Geometric Simplex Features**: k-simplex vertex computations from Cantor measure
+- **Adaptive Augmentation**: Progressive augmentation escalation to prevent overfitting
+- **Beatrix Formula Suite**: Flow alignment, hierarchical coherence, and multi-scale consistency losses
+
+### Adaptive Augmentation System
+
+The trainer includes an intelligent augmentation system that monitors train/validation accuracy gap and progressively enables more augmentation:
+
+1. **Baseline**: RandomCrop + RandomHorizontalFlip
+2. **Stage 1**: + ColorJitter
+3. **Stage 2**: + RandomRotation  
+4. **Stage 3**: + RandomAffine
+5. **Stage 4**: + RandomErasing
+6. **Stage 5**: + AutoAugment (CIFAR policy)
+7. **Stage 6**: Enable Mixup (α=0.2)
+8. **Stage 7**: Enable CutMix (α=1.0) - Final stage
+
+When train accuracy exceeds validation accuracy by 2% or more, the system automatically escalates to the next augmentation stage.
+
 ## Available Models (Best Checkpoints Only)
 
 | Model Name | Training Session | Accuracy | Epoch | Weights Path | Logs Path |
@@ -21,9 +43,10 @@ with geometric simplex features for vision tasks.
 | beatrix-cifar100 | 20251007_182851 | 0.5819 | 42 | `weights/beatrix-cifar100/20251007_182851` | `N/A` |
 | beatrix-cifar100 | 20251007_195812 | 0.4701 | 42 | `weights/beatrix-cifar100/20251007_195812` | `logs/beatrix-cifar100/20251007_195812` |
 | beatrix-cifar100 | 20251007_193112 | 0.2802 | 10 | `weights/beatrix-cifar100/20251007_193112` | `N/A` |
+| beatrix-cifar100 | 20251007_203741 | 0.0470 | 0 | `weights/beatrix-cifar100/20251007_203741` | `logs/beatrix-cifar100/20251007_203741` |
 
 
-## Latest Updated Model: beatrix-cifar100 (Session: 20251007_195812)
+## Latest Updated Model: beatrix-cifar100 (Session: 20251007_203741)
 
 ### Model Details
 
@@ -37,11 +60,12 @@ with geometric simplex features for vision tasks.
 
 ### Training Details
 
-- **Training Session**: 20251007_195812
-- **Best Accuracy**: 0.4701
-- **Epochs Trained**: 42
+- **Training Session**: 20251007_203741
+- **Best Accuracy**: 0.0470
+- **Epochs Trained**: 0
 - **Batch Size**: 512
 - **Learning Rate**: 0.0001
+- **Adaptive Augmentation**: Enabled
 
 ### Loss Configuration
 
@@ -54,7 +78,7 @@ with geometric simplex features for vision tasks.
 
 Training logs are available in the repository at:
 ```
-logs/beatrix-cifar100/20251007_195812
+logs/beatrix-cifar100/20251007_203741
 ```
 
 To view them locally:
@@ -63,11 +87,32 @@ To view them locally:
 git clone https://huggingface.co/AbstractPhil/vit-beatrix
 
 # View logs in TensorBoard
-tensorboard --logdir vit-beatrix/logs/beatrix-cifar100/20251007_195812
+tensorboard --logdir vit-beatrix/logs/beatrix-cifar100/20251007_203741
 ```
 
 ## Usage
 
+### Installation
+
+For Google Colab:
+```python
+# Install for Colab
+try:
+  !pip uninstall -qy geometricvocab
+except:
+  pass
+
+!pip install -q git+https://github.com/AbstractEyes/lattice_vocabulary.git
+```
+
+For local environments:
+```bash
+# install the repo into your environment
+pip install -q git+https://github.com/AbstractEyes/lattice_vocabulary.git
+```
+
+### Loading Models
+
 ```python
 from geovocab2.train.model.vit_beatrix import SimplifiedGeometricClassifier
 from safetensors.torch import load_file
@@ -90,7 +135,7 @@ for key, info in sorted(manifest.items(), key=lambda x: x[1]['accuracy'], revers
 # Download weights for the latest training session of beatrix-cifar100
 weights_path = hf_hub_download(
     repo_id="AbstractPhil/vit-beatrix",
-    filename="weights/beatrix-cifar100/20251007_195812/model.safetensors"
+    filename="weights/beatrix-cifar100/20251007_203741/model.safetensors"
 )
 
 # Load model

logs/beatrix-cifar100/20251007_203741/events.out.tfevents.1759869463.f41d735e1a7f.65607.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ca3589705bba9a0c2248a76818c50adfacecc9ef7a63f9a9539c3babe019b84
+size 88

manifest.json

@@ -31,5 +31,17 @@
     "depth": 12,
     "batch_size": 512,
     "learning_rate": 0.0001
+  },
+  "beatrix-cifar100_20251007_203741": {
+    "accuracy": 0.04703010115772486,
+    "epoch": 0,
+    "timestamp": "20251007_203741",
+    "model_name": "beatrix-cifar100",
+    "path": "weights/beatrix-cifar100/20251007_203741",
+    "logs_path": "logs/beatrix-cifar100/20251007_203741",
+    "embed_dim": 256,
+    "depth": 12,
+    "batch_size": 512,
+    "learning_rate": 0.0001
   }
 }

weights/beatrix-cifar100/20251007_203741/config.json

@@ -0,0 +1,45 @@
+{
+  "num_classes": 100,
+  "img_size": 32,
+  "patch_size": 4,
+  "embed_dim": 256,
+  "k_simplex": 4,
+  "depth": 12,
+  "num_heads": 8,
+  "mlp_ratio": 4.0,
+  "dropout": 0.0,
+  "pe_levels": 12,
+  "pe_features_per_level": 2,
+  "pe_smooth_tau": 0.25,
+  "simplex_feature_weight": 0.5,
+  "batch_size": 512,
+  "num_epochs": 50,
+  "learning_rate": 0.0001,
+  "weight_decay": 0.005,
+  "warmup_epochs": 10,
+  "task_loss_weight": 0.5,
+  "flow_loss_weight": 0.5,
+  "coherence_loss_weight": 0.3,
+  "multiscale_loss_weight": 0.2,
+  "volume_reg_weight": 0.1,
+  "use_adaptive_augmentation": true,
+  "overfit_threshold": 0.02,
+  "mixup_alpha": 0.2,
+  "cutmix_alpha": 1.0,
+  "device": "cuda",
+  "num_workers": 4,
+  "pin_memory": true,
+  "save_dir": "./checkpoints",
+  "save_every": 10,
+  "use_safetensors": true,
+  "timestamp_dirs": true,
+  "push_to_hub": true,
+  "hub_model_id": "AbstractPhil/vit-beatrix",
+  "hub_model_name": "beatrix-cifar100",
+  "hub_upload_best_only": true,
+  "hub_upload_every_n_epochs": 10,
+  "use_tensorboard": true,
+  "log_dir": "./logs",
+  "log_every": 50,
+  "train_baseline": false
+}

weights/beatrix-cifar100/20251007_203741/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e46631fad61dd7418830b1b01b1266024244b295b08853e8a7098ea683b91f6d
+size 38116128