[Metrics] Fix minor inconsistencies in bucket progression (vllm-project#17262)

Signed-off-by: DarkLight1337 <[email protected]>
Signed-off-by: Mu Huai <[email protected]>

Author: DarkLight1337

vllm/engine/metrics.py

@@ -145,7 +145,7 @@ def __init__(self, labelnames: List[str], vllm_config: VllmConfig):
             documentation="Histogram of number of tokens per engine_step.",
             labelnames=labelnames,
             buckets=[
-                1, 8, 16, 32, 64, 128, 256, 512, 1024, 2048, 4096, 8096, 16192
+                1, 8, 16, 32, 64, 128, 256, 512, 1024, 2048, 4096, 8192, 16384
             ])
         self.histogram_time_to_first_token = self._histogram_cls(
             name="vllm:time_to_first_token_seconds",

vllm/v1/metrics/loggers.py

@@ -233,8 +233,8 @@ def __init__(self, vllm_config: VllmConfig, engine_index: int = 0):
                 name="vllm:iteration_tokens_total",
                 documentation="Histogram of number of tokens per engine_step.",
                 buckets=[
-                    1, 8, 16, 32, 64, 128, 256, 512, 1024, 2048, 4096, 8096,
-                    16192
+                    1, 8, 16, 32, 64, 128, 256, 512, 1024, 2048, 4096, 8192,
+                    16384
                 ],
                 labelnames=labelnames).labels(*labelvalues)