Remove best of metrics to align with latest vllm

triton-inference-server · Nov 1, 2024 · 9897102 · 9897102
1 parent 4a8b737
commit 9897102
Showing 1 changed file with 0 additions and 10 deletions.
diff --git a/src/utils/metrics.py b/src/utils/metrics.py
@@ -76,11 +76,6 @@ def __init__(self, labels: List[str], max_model_len: int):
             description="Number of generation tokens processed.",
             kind=pb_utils.MetricFamily.HISTOGRAM,
         )
-        self.histogram_best_of_request_family = pb_utils.MetricFamily(
-            name="vllm:request_params_best_of",
-            description="Histogram of the best_of request parameter.",
-            kind=pb_utils.MetricFamily.HISTOGRAM,
-        )
         self.histogram_n_request_family = pb_utils.MetricFamily(
             name="vllm:request_params_n",
             description="Histogram of the n request parameter.",
@@ -159,10 +154,6 @@ def __init__(self, labels: List[str], max_model_len: int):
                 buckets=build_1_2_5_buckets(max_model_len),
             )
         )
-        self.histogram_best_of_request = self.histogram_best_of_request_family.Metric(
-            labels=labels,
-            buckets=[1, 2, 5, 10, 20],
-        )
         self.histogram_n_request = self.histogram_n_request_family.Metric(
             labels=labels,
             buckets=[1, 2, 5, 10, 20],
@@ -247,7 +238,6 @@ def log(self, stats: VllmStats) -> None:
                 self.metrics.histogram_num_generation_tokens_request,
                 stats.num_generation_tokens_requests,
             ),
-            (self.metrics.histogram_best_of_request, stats.best_of_requests),
             (self.metrics.histogram_n_request, stats.n_requests),
         ]