test

Author: xzrderek

eval_protocol/benchmarks/test_aime25.py

@@ -86,10 +86,10 @@ def aime2025_dataset_adapter(rows: List[Dict[str, Any]]) -> List[EvaluationRow]:
     dataset_adapter=aime2025_dataset_adapter,
     completion_params=[
         {
-            "max_tokens": 131000,
-            "extra_body": {"reasoning_effort": "low"},
-            # "model": "fireworks_ai/accounts/pyroworks/deployedModels/glm-4p6-qpwrimne",
-            "model": "fireworks_ai/accounts/fireworks/models/gpt-oss-120b",
+            "max_tokens": 65536,
+            # "extra_body": {"reasoning_effort": "low"},
+            "model": "fireworks_ai/accounts/pyroworks/deployedModels/glm-4p6-qpwrimne",
+            # "model": "fireworks_ai/accounts/fireworks/models/gpt-oss-120b",
         }
     ],
     rollout_processor=SingleTurnRolloutProcessor(),

eval_protocol/pytest/default_single_turn_rollout_process.py

@@ -65,9 +65,12 @@ async def process_row(row: EvaluationRow) -> EvaluationRow:
                 request_params["tools"] = row.tools
 
             chunks = []
+            print("time: ", time.time())
 
             stream = await acompletion(**request_params)
             async for chunk in stream:
+                print("chunk added at time: ", time.time())
+                print("chunk: ", chunk)
                 chunks.append(chunk)
 
             response = litellm.stream_chunk_builder(chunks, messages_payload)