Add sum to usage statistics

browseruse_bench/utils/stats.py

@@ -1,6 +1,7 @@
 """Statistical utility functions."""
 from __future__ import annotations
 
+from math import fsum
 from typing import Any, Dict, List, Optional
 
 
@@ -98,7 +99,9 @@ def calculate_usage_stats(tasks: List[Dict[str, Any]], path: str = "evaluation_d
                     vals.append(float(usage[field]))
 
         if vals:
-            usage_stats[field] = _calc_stats(vals)
+            field_stats = _calc_stats(vals)
+            field_stats["sum"] = fsum(vals)
+            usage_stats[field] = field_stats
 
     return usage_stats
 

tests/browseruse_bench/test_stats.py

@@ -33,3 +33,48 @@ def test_filter_empty_list(self):
         """Test filtering empty list returns empty."""
         result = filter_tasks_by_label([], key="predicted_label", val=1)
         assert result == []
+
+
+class TestGenerateEvaluationSummary:
+    """Tests for generated summary statistics."""
+
+    def test_usage_statistics_include_sum(self):
+        """Usage fields should expose a total across evaluated tasks."""
+        results = [
+            {
+                "task_id": "1",
+                "predicted_label": 1,
+                "evaluation_details": {
+                    "agent_metrics": {
+                        "usage": {
+                            "total_prompt_tokens": 100,
+                            "total_completion_tokens": 20,
+                            "total_tokens": 120,
+                            "entry_count": 1,
+                        }
+                    }
+                },
+            },
+            {
+                "task_id": "2",
+                "predicted_label": 0,
+                "evaluation_details": {
+                    "agent_metrics": {
+                        "usage": {
+                            "total_prompt_tokens": 200,
+                            "total_completion_tokens": 80,
+                            "total_tokens": 280,
+                            "entry_count": 2,
+                        }
+                    }
+                },
+            },
+        ]
+
+        summary = generate_evaluation_summary(results, total=2)
+
+        usage_stats = summary["metrics_statistics"]["usage"]
+        assert usage_stats["total_prompt_tokens"]["sum"] == 300
+        assert usage_stats["total_completion_tokens"]["sum"] == 100
+        assert usage_stats["total_tokens"]["sum"] == 400
+        assert usage_stats["entry_count"]["sum"] == 3