fix(paperbench): handle uneven seed counts in run parsing

project/paperbench/paperbench/metrics.py

@@ -293,7 +293,7 @@ def parse_entry_by_format(
             eval_run = EvaluationRun(seed=seed, paper_evaluations={})
             for data in filtered_paper_data.values():
                 # some evaluation runs may not have all papers evaluated
-                if seed == len(data):
+                if seed >= len(data):
                     continue
                 paper_eval = data[seed]["paper_eval"]
                 eval_run.paper_evaluations[paper_eval.paper_id] = paper_eval

project/paperbench/tests/unit/test_metrics.py

@@ -0,0 +1,52 @@
+import json
+
+from paperbench.metrics import parse_run_data
+
+
+def _graded_task_tree() -> dict:
+    return {
+        "id": "task",
+        "requirements": "test requirement",
+        "weight": 1,
+        "sub_tasks": [],
+        "task_category": "Code Development",
+        "score": 1.0,
+        "valid_score": True,
+        "explanation": "ok",
+    }
+
+
+def _entry(paper_id: str, run_id: str, timestamp: str) -> dict:
+    return {
+        "record_type": "extra",
+        "timestamp": timestamp,
+        "data": {
+            "run_group_id": "group_agent",
+            "run_id": run_id,
+            "pb_result": {
+                "paperbench_result": {
+                    "paper_id": paper_id,
+                    "judge_output": {"graded_task_tree": _graded_task_tree()},
+                }
+            },
+        },
+    }
+
+
+def test_parse_run_data_handles_papers_with_uneven_seed_counts(tmp_path):
+    entries = [
+        _entry("paper-a", "a-1", "2026-01-03T00:00:00Z"),
+        _entry("paper-a", "a-2", "2026-01-02T00:00:00Z"),
+        _entry("paper-a", "a-3", "2026-01-01T00:00:00Z"),
+        _entry("paper-b", "b-1", "2026-01-03T00:00:00Z"),
+    ]
+    run_file = tmp_path / "runs.jsonl"
+    run_file.write_text("".join(f"{json.dumps(entry)}\n" for entry in entries))
+
+    runs = parse_run_data(tmp_path)
+
+    assert list(runs) == ["agent"]
+    assert len(runs["agent"]) == 3
+    assert set(runs["agent"][0].paper_evaluations) == {"paper-a", "paper-b"}
+    assert set(runs["agent"][1].paper_evaluations) == {"paper-a"}
+    assert set(runs["agent"][2].paper_evaluations) == {"paper-a"}