Skip to content

GitLab

"tests/python/vscode:/vscode.git/clone" did not exist on "af8451c2235939be33f55896bc81122d27795a93"
Unverified Commit 26ebb849 authored by fzyzcjy's avatar fzyzcjy Committed by GitHub
Browse files

Tiny refactor bench_serving to extract RequestFuncOutput.init_new (#6108)

parent 02973cd9
Hide whitespace changes
Inline Side-by-side
Showing with 10 additions and 8 deletions
python/sglang/bench_serving.py
View file @ 26ebb849
...@@ -73,6 +73,12 @@ class RequestFuncOutput: ...@@ -73,6 +73,12 @@ class RequestFuncOutput:
error: str = "" error: str = ""
output_len: int = 0 output_len: int = 0
@staticmethod
def init_new(request_func_input: RequestFuncInput):
output = RequestFuncOutput()
output.prompt_len = request_func_input.prompt_len
return output
def remove_prefix(text: str, prefix: str) -> str: def remove_prefix(text: str, prefix: str) -> str:
return text[len(prefix) :] if text.startswith(prefix) else text return text[len(prefix) :] if text.startswith(prefix) else text
...@@ -114,8 +120,7 @@ async def async_request_trt_llm( ...@@ -114,8 +120,7 @@ async def async_request_trt_llm(
if args.disable_ignore_eos: if args.disable_ignore_eos:
del payload["min_length"] del payload["min_length"]
del payload["end_id"] del payload["end_id"]
output = RequestFuncOutput() output = RequestFuncOutput.init_new(request_func_input)
output.prompt_len = request_func_input.prompt_len
ttft = 0.0 ttft = 0.0
st = time.perf_counter() st = time.perf_counter()
...@@ -186,8 +191,7 @@ async def async_request_openai_completions( ...@@ -186,8 +191,7 @@ async def async_request_openai_completions(
} }
headers = get_auth_headers() headers = get_auth_headers()
output = RequestFuncOutput() output = RequestFuncOutput.init_new(request_func_input)
output.prompt_len = request_func_input.prompt_len
generated_text = "" generated_text = ""
output_len = request_func_input.output_len output_len = request_func_input.output_len
...@@ -269,8 +273,7 @@ async def async_request_truss( ...@@ -269,8 +273,7 @@ async def async_request_truss(
} }
headers = get_auth_headers() headers = get_auth_headers()
output = RequestFuncOutput() output = RequestFuncOutput.init_new(request_func_input)
output.prompt_len = request_func_input.prompt_len
generated_text = "" generated_text = ""
ttft = 0.0 ttft = 0.0
...@@ -355,8 +358,7 @@ async def async_request_sglang_generate( ...@@ -355,8 +358,7 @@ async def async_request_sglang_generate(
headers = get_auth_headers() headers = get_auth_headers()
output = RequestFuncOutput() output = RequestFuncOutput.init_new(request_func_input)
output.prompt_len = request_func_input.prompt_len
generated_text = "" generated_text = ""
output_len = request_func_input.output_len output_len = request_func_input.output_len
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment