Update bench_serving.py (#4454)
This commit is contained in:
@@ -128,7 +128,7 @@ async def async_request_trt_llm(
|
|||||||
timestamp = time.perf_counter()
|
timestamp = time.perf_counter()
|
||||||
# First token
|
# First token
|
||||||
if ttft == 0.0:
|
if ttft == 0.0:
|
||||||
ttft = time.perf_counter() - st
|
ttft = timestamp - st
|
||||||
output.ttft = ttft
|
output.ttft = ttft
|
||||||
|
|
||||||
# Decoding phase
|
# Decoding phase
|
||||||
|
|||||||
Reference in New Issue
Block a user