Longer time for A100

triton-inference-server · Aug 15, 2024 · 993f0c7 · 993f0c7
1 parent 11df957
commit 993f0c7
Showing 1 changed file with 4 additions and 4 deletions.
diff --git a/ci/L0_backend_vllm/metrics_test/vllm_metrics_test.py b/ci/L0_backend_vllm/metrics_test/vllm_metrics_test.py
@@ -127,14 +127,14 @@ def test_vllm_metrics(self):
 
         # vllm:time_to_first_token_seconds
         self.assertEqual(metrics_dict["vllm:time_to_first_token_seconds_count"], 3)
-        self.assertTrue(
-            0 < metrics_dict["vllm:time_to_first_token_seconds_sum"] < 0.0005
-        )
+        self.assertTrue(0 < metrics_dict["vllm:time_to_first_token_seconds_sum"] < 0.01)
+        self.assertEqual(metrics_dict["vllm:time_to_first_token_seconds_bucket"], 3)
         # vllm:time_per_output_token_seconds
         self.assertEqual(metrics_dict["vllm:time_per_output_token_seconds_count"], 45)
         self.assertTrue(
-            0 <= metrics_dict["vllm:time_per_output_token_seconds_sum"] <= 0.005
+            0 < metrics_dict["vllm:time_per_output_token_seconds_sum"] < 0.1
         )
+        self.assertEqual(metrics_dict["vllm:time_per_output_token_seconds_bucket"], 45)
 
     def test_vllm_metrics_disabled(self):
         # Test vLLM metrics