Upload model_metadata.json with huggingface_hub

model_metadata.json

@@ -1,14 +1,18 @@
 {
-  "version": "clean-baseline-v1.0",
+  "version": "clean-baseline-v2.0-hf",
   "model": "vit_base_patch16_224",
-  "num_classes": 18,
+  "num_classes": 23,
   "class_names": [
     "call",
     "dislike",
     "fist",
     "four",
+    "grabbing",
+    "grip",
     "like",
+    "middle_finger",
     "mute",
+    "no_gesture",
     "ok",
     "one",
     "palm",
@@ -19,22 +23,25 @@
     "stop_inverted",
     "three",
     "three2",
+    "three3",
     "two_up",
     "two_up_inverted"
   ],
   "img_size": 224,
-  "test_accuracy": 0.8327051926298158,
-  "confusion_matrix_diagonality": 83.27051926298158,
+  "test_accuracy": 0.9248953459738981,
+  "confusion_matrix_diagonality": 92.4895345973898,
   "best_epoch": 15,
-  "best_val_acc": 82.55131964809384,
-  "best_val_f1": 0.8329613461728465,
+  "best_val_acc": 89.87750372454892,
+  "best_val_f1": 0.9248777400191369,
   "train_acc": 100.0,
   "dataset": {
-    "name": "HaGRID Sample 30K (384p)",
-    "source": "https://www.kaggle.com/datasets/innominate817/hagrid-sample-30k-384p",
-    "train_size": 22283,
-    "val_size": 4774,
-    "test_size": 4776
+    "name": "HaGRID Subset",
+    "source": "local",
+    "path": "data/hagrid_subset",
+    "train_size": 103171,
+    "val_size": 12082,
+    "test_size": 20305,
+    "splits": "fixed"
   },
   "training_config": {
     "batch_size": 32,
@@ -59,6 +66,7 @@
       0.225
     ],
     "augmentation": "none",
+    "rationale": "augmentation corrupts gesture semantics",
     "tta": false
   },
   "architecture": {
@@ -67,5 +75,5 @@
     "hidden_dim": null,
     "dropout": 0.0
   },
-  "timestamp": "2026-01-09T23:47:05.506496"
+  "timestamp": "2026-01-14T22:44:08.208470"
 }