Upload folder using huggingface_hub

eval_results/comparison_report_base_fullft.json

@@ -0,0 +1,74 @@
+{
+  "timestamp": "2026-03-09 08:34:31",
+  "scoring_method": "PhyX-aligned (DeepSeek-V3 judge, 5-shot ICE, 5 retries)",
+  "base": {
+    "model": "Base (Qwen2.5-VL-3B-Instruct)",
+    "total": 1533,
+    "string_matches": 5,
+    "llm_calls": 1528,
+    "llm_matches": 278,
+    "final_correct": 283,
+    "final_acc": 18.46,
+    "category_stats": {
+      "Mechanics": {
+        "total": 276,
+        "correct": 65
+      },
+      "Waves/Acoustics": {
+        "total": 253,
+        "correct": 35
+      },
+      "Electromagnetism": {
+        "total": 275,
+        "correct": 51
+      },
+      "Modern Physics": {
+        "total": 222,
+        "correct": 54
+      },
+      "Optics": {
+        "total": 252,
+        "correct": 44
+      },
+      "Thermodynamics": {
+        "total": 255,
+        "correct": 34
+      }
+    }
+  },
+  "fullft": {
+    "model": "SFT-fullft (Cold-Start)",
+    "total": 1533,
+    "string_matches": 13,
+    "llm_calls": 1520,
+    "llm_matches": 331,
+    "final_correct": 344,
+    "final_acc": 22.44,
+    "category_stats": {
+      "Mechanics": {
+        "total": 276,
+        "correct": 77
+      },
+      "Waves/Acoustics": {
+        "total": 253,
+        "correct": 42
+      },
+      "Electromagnetism": {
+        "total": 275,
+        "correct": 62
+      },
+      "Modern Physics": {
+        "total": 222,
+        "correct": 63
+      },
+      "Optics": {
+        "total": 252,
+        "correct": 55
+      },
+      "Thermodynamics": {
+        "total": 255,
+        "correct": 45
+      }
+    }
+  }
+}