Update beatrix-dualstream-base (Epoch 0, Acc: 0.0816)

README.md

@@ -10,25 +10,25 @@ license: mit
 
 # ViT-Beatrix Dual-Stream Family
 
-## Current Experiment: beatrix-trainD-pretrained-geometry
+## Current Experiment: beatrix-dualstream-base
 
-**Model Path**: `weights/beatrix-trainD-pretrained-geometry/20251009_001904_frozen_geometry/`
+**Model Path**: `weights/beatrix-dualstream-base/20251009_011847/`
 
 
 ## Architecture
 
 - **Visual Dimension**: 512
-- **Geometric Dimension**: 256
-- **Geometric Tokens**: 8
-- **Dual Blocks**: 8 layers
-- **k-simplex**: 4
+- **Geometric Dimension**: 512
+- **Geometric Tokens**: 16
+- **Dual Blocks**: 16 layers
+- **k-simplex**: 12
 
 ## Performance
 
-- **Best Accuracy**: 0.3331
-- **Current Epoch**: 10
+- **Best Accuracy**: 0.0816
+- **Current Epoch**: 0
 - **Dataset**: CIFAR-100
 
 ---
 
-*Last updated: Epoch 10 | Best Accuracy: 0.3331*
+*Last updated: Epoch 0 | Best Accuracy: 0.0816*

weights/beatrix-dualstream-base/20251009_011847/config.json

@@ -0,0 +1,83 @@
+{
+  "num_classes": 100,
+  "img_size": 32,
+  "patch_size": 4,
+  "visual_dim": 512,
+  "geom_dim": 512,
+  "k_simplex": 12,
+  "depth": 16,
+  "num_heads": 8,
+  "mlp_ratio": 4.0,
+  "dropout": 0.0,
+  "num_geom_tokens": 16,
+  "freeze_geometric_stream": false,
+  "geometric_checkpoint_path": null,
+  "pe_levels": 24,
+  "pe_features_per_level": 2,
+  "pe_smooth_tau": 0.25,
+  "simplex_init_method": "regular",
+  "simplex_init_scale": 1.0,
+  "batch_size": 512,
+  "num_epochs": 100,
+  "learning_rate": 0.0001,
+  "weight_decay": 0.005,
+  "warmup_epochs": 10,
+  "task_loss_weight": 0.5,
+  "flow_loss_weight": 1.0,
+  "coherence_loss_weight": 0.3,
+  "multiscale_loss_weight": 0.2,
+  "use_adaptive_augmentation": false,
+  "overfit_threshold": 0.05,
+  "augmentation_cooldown_epochs": 5,
+  "min_accuracy_for_augmentation": 0.45,
+  "mixup_alpha": 0.2,
+  "cutmix_alpha": 1.0,
+  "use_cutmix_schedule": false,
+  "cutmix_schedule": [
+    [
+      0,
+      0.2
+    ],
+    [
+      20,
+      0.5
+    ],
+    [
+      40,
+      1.0
+    ],
+    [
+      60,
+      1.2
+    ],
+    [
+      80,
+      1.5
+    ],
+    [
+      100,
+      1.8
+    ],
+    [
+      120,
+      2.0
+    ]
+  ],
+  "device": "cuda",
+  "num_workers": 4,
+  "pin_memory": true,
+  "save_dir": "./checkpoints_dualstream",
+  "save_every": 10,
+  "use_safetensors": true,
+  "timestamp_dirs": true,
+  "push_to_hub": true,
+  "hub_model_id": "AbstractPhil/vit-beatrix-dualstream",
+  "hub_model_name": "beatrix-dualstream-base",
+  "hub_upload_best_only": true,
+  "hub_upload_every_n_epochs": 10,
+  "use_tensorboard": true,
+  "log_dir": "./logs_dualstream",
+  "log_every": 50,
+  "monitor_stream_health": true,
+  "log_stream_norms": true
+}

weights/beatrix-dualstream-base/20251009_011847/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:42536611e612c2a8f345a107bc036e88fc4a6e812c5d3d402947063695fc1314
+size 538890112