Add results file for gemini-2.5-flash-preview-04-17

results/google/gemini-2.5-flash-preview-04-17_results_2025-05-11_12-29-53.json

@@ -0,0 +1,158 @@
+{
+  "results": {
+    "average_score": 8.970212765957447,
+    "speed": 9.621688633480778,
+    "contamination_score": 0,
+    "execution_time": 2873.300213,
+    "errors": [
+      {
+        "error": "500 INTERNAL. {'error': {'code': 500, 'message': 'An internal error has occurred. Please retry or report in https://developers.generativeai.google/guide/troubleshooting', 'status': 'INTERNAL'}}",
+        "prompt": "{'role': 'user', 'content': \"You are a function calling AI model. You are provided with function signatures within <tools> </tools> XML tags. You may call one or more functions to assist with the user query. Don't make assumptions about what values to plug into functions\\n<tools>\\n[\\n  {\\n    'tool': 'get_current_weather',\\n    'description': 'Gets the current weather conditions for a given city.',\\n    'parameters': {\\n      'city': {\\n        'type': 'string',\\n        'required': true\\n      },\\n      'unit': {\\n        'type': 'string',\\n        'required': false,\\n        'valid_values': ['celsius', 'fahrenheit'],\\n        'default': 'celsius'\\n      }\\n    }\\n  },\\n  {\\n    'tool': 'calculate_distance',\\n    'description': 'Calculates the driving distance between two cities.',\\n    'parameters': {\\n      'city1': {\\n        'type': 'string',\\n        'required': true\\n      },\\n      'city2': {\\n        'type': 'string',\\n        'required': true\\n      }\\n    }\\n  },\\n  {\\n    'tool': 'send_email',\\n    'description': 'Sends an email to a specified recipient.',\\n    'parameters': {\\n      'to': {\\n        'type': 'string',\\n        'required': true\\n      },\\n      'subject': {\\n        'type': 'string',\\n        'required': true\\n      },\\n      'body': {\\n        'type': 'string',\\n        'required': true\\n      }\\n    }\\n  }\\n]\\n</tools>\\n\\nFor each function call return a json object with function name and arguments within <tool_call> </tool_call> tags with the following schema:\\n<tool_call>\\n{'arguments': <args-dict>, 'name': <function-name>}\\n</tool_call>\\n\\n\\nQuestion:\\n\u0645\u0627 \u0647\u0648 \u0627\u0644\u0637\u0642\u0633 \u0641\u064a \u0627\u0644\u0642\u0627\u0647\u0631\u0629 \u0627\u0644\u064a\u0648\u0645\u061f\"}"
+      }
+    ],
+    "scores_by_category": [
+      {
+        "category": "Sentiment Analysis",
+        "average_score": 10.0,
+        "count": 9
+      },
+      {
+        "category": "Hallucination",
+        "average_score": 10.0,
+        "count": 3
+      },
+      {
+        "category": "Coding",
+        "average_score": 10.0,
+        "count": 3
+      },
+      {
+        "category": "Reading Comprehension",
+        "average_score": 10.0,
+        "count": 17
+      },
+      {
+        "category": "Diacritization",
+        "average_score": 9.916666666666666,
+        "count": 12
+      },
+      {
+        "category": "MMLU",
+        "average_score": 9.50413223140496,
+        "count": 121
+      },
+      {
+        "category": "Long Context",
+        "average_score": 9.5,
+        "count": 4
+      },
+      {
+        "category": "Dialect Detection",
+        "average_score": 9.363636363636363,
+        "count": 11
+      },
+      {
+        "category": "RAG QA",
+        "average_score": 9.317073170731707,
+        "count": 41
+      },
+      {
+        "category": "Instruction Following",
+        "average_score": 9.285714285714286,
+        "count": 7
+      },
+      {
+        "category": "General Knowledge",
+        "average_score": 9.206349206349206,
+        "count": 63
+      },
+      {
+        "category": "Reasoning & Math",
+        "average_score": 9.069767441860465,
+        "count": 43
+      },
+      {
+        "category": "Trust & Safety",
+        "average_score": 9.0,
+        "count": 30
+      },
+      {
+        "category": "Entity Extraction",
+        "average_score": 8.8,
+        "count": 5
+      },
+      {
+        "category": "Arabic Language & Grammar",
+        "average_score": 8.352941176470589,
+        "count": 17
+      },
+      {
+        "category": "Writing (incl Dialects)",
+        "average_score": 7.636363636363637,
+        "count": 22
+      },
+      {
+        "category": "Summarization",
+        "average_score": 7.5,
+        "count": 8
+      },
+      {
+        "category": "Transliteration",
+        "average_score": 7.5,
+        "count": 6
+      },
+      {
+        "category": "Translation (incl Dialects)",
+        "average_score": 7.444444444444445,
+        "count": 36
+      },
+      {
+        "category": "Paraphrasing",
+        "average_score": 6.666666666666667,
+        "count": 6
+      },
+      {
+        "category": "Structuring",
+        "average_score": 5.666666666666667,
+        "count": 3
+      },
+      {
+        "category": "Function Calling",
+        "average_score": 5.0,
+        "count": 3
+      }
+    ],
+    "scores_by_format": [
+      {
+        "format": "Fill-in-the-blank",
+        "average_score": 10.0,
+        "count": 8
+      },
+      {
+        "format": "Short Answer",
+        "average_score": 10.0,
+        "count": 5
+      },
+      {
+        "format": "MCQ",
+        "average_score": 9.292576419213974,
+        "count": 229
+      },
+      {
+        "format": "Generation",
+        "average_score": 8.587719298245615,
+        "count": 228
+      }
+    ]
+  },
+  "config": {
+    "model": "google/gemini-2.5-flash-preview-04-17",
+    "model_sha": "na",
+    "submitted_time": "2025-05-11 11:41:55",
+    "likes": -1,
+    "params": 999,
+    "license": "closed",
+    "model_source": "API",
+    "model_category": "Large"
+  }
+}