same as aime now

Author: xzrderek

eval_protocol/quickstart/llm_judge.py

@@ -39,34 +39,30 @@ def aime2025_dataset_adapter(rows: List[Dict[str, Any]]) -> List[EvaluationRow]:
     return converted
 
 
-@pytest.mark.asyncio
 @evaluation_test(
     input_dataset=[
         "https://huggingface.co/datasets/opencompass/AIME2025/raw/main/aime2025-I.jsonl",
         "https://huggingface.co/datasets/opencompass/AIME2025/raw/main/aime2025-II.jsonl",
     ],
     dataset_adapter=aime2025_dataset_adapter,
     completion_params=[
-        # {
-        #     "model": "fireworks_ai/accounts/fireworks/models/qwen3-235b-a22b-instruct-2507",
-        # },
-        {"model": "gpt-4.1"},
         {
             "max_tokens": 131000,
-            "extra_body": {"reasoning_effort": "medium"},
+            "extra_body": {"reasoning_effort": "low"},
             "model": "fireworks_ai/accounts/fireworks/models/gpt-oss-120b",
         },
         {
             "max_tokens": 131000,
-            "extra_body": {"reasoning_effort": "low"},
-            "model": "fireworks_ai/accounts/fireworks/models/gpt-oss-20b",
+            "extra_body": {"reasoning_effort": "medium"},
+            "model": "fireworks_ai/accounts/fireworks/models/gpt-oss-120b",
         },
     ],
     rollout_processor=SingleTurnRolloutProcessor(),
-    # preprocess_fn=split_multi_turn_rows,
+    aggregation_method="mean",
+    passed_threshold=0.8,
     num_runs=1,
     max_dataset_rows=1,
-    max_concurrent_rollouts=64,
+    max_concurrent_rollouts=4,
     mode="pointwise",
 )
 async def test_llm_judge(row: EvaluationRow) -> EvaluationRow: