update

Author: xzrderek

eval_protocol/quickstart/llm_judge_braintrust.py

@@ -51,7 +51,7 @@
     input_rows=input_rows,
     rollout_processor=SingleTurnRolloutProcessor(),
     preprocess_fn=multi_turn_assistant_to_ground_truth,
-    max_concurrent_rollouts=64,
+    max_concurrent_evaluations=2,
 )
 async def test_llm_judge(row: EvaluationRow) -> EvaluationRow:
     return await aha_judge(row)

eval_protocol/quickstart/llm_judge_langfuse.py

@@ -51,8 +51,7 @@
     input_rows=[input_rows],
     rollout_processor=SingleTurnRolloutProcessor(),
     preprocess_fn=multi_turn_assistant_to_ground_truth,
-    max_concurrent_rollouts=64,
-    max_concurrent_evaluations=16,
+    max_concurrent_evaluations=2,
 )
 async def test_llm_judge(row: EvaluationRow) -> EvaluationRow:
     return await aha_judge(row)

eval_protocol/quickstart/llm_judge_langsmith.py

@@ -76,6 +76,7 @@ def fetch_langsmith_traces_as_evaluation_rows(
     input_rows=[input_rows],
     rollout_processor=SingleTurnRolloutProcessor(),
     preprocess_fn=multi_turn_assistant_to_ground_truth,
+    max_concurrent_evaluations=2,
 )
 async def test_llm_judge_langsmith(row: EvaluationRow) -> EvaluationRow:
     """LLM Judge evaluation over LangSmith-sourced rows, persisted locally by Eval Protocol.

eval_protocol/quickstart/llm_judge_openai_responses.py

@@ -57,6 +57,7 @@
     input_rows=[input_rows],
     rollout_processor=SingleTurnRolloutProcessor(),
     preprocess_fn=multi_turn_assistant_to_ground_truth,
+    max_concurrent_evaluations=2,
 )
 async def test_llm_judge_openai_responses(row: EvaluationRow) -> EvaluationRow:
     return await aha_judge(row)