chenchenghao/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 11 Packages Projects Releases Wiki Activity

[Misc] Normalize --host parameter to use plain hostname without scheme (#19309)

Browse Source 
Co-authored-by: 墨楼 <huangzhilin.hzl@antgroup.com>
Co-authored-by: Liangsheng Yin <lsyincs@gmail.com>
Co-authored-by: Liangsheng Yin <hnyls2002@gmail.com>
This commit is contained in:
Julian Huang
2026-02-25 16:37:24 +08:00
committed by GitHub
parent f75abb4521
commit a55f658835
8 changed files with 52 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
examples/runtime/multimodal/llama3_llava_server.py
View File

@@ -21,6 +21,8 @@ import aiohttp
import requests
from llava.conversation import conv_llava_llama_3
from sglang.utils import normalize_base_url
async def send_request(url, data, delay=0):
await asyncio.sleep(delay)
@@ -31,7 +33,7 @@ async def send_request(url, data, delay=0):
async def test_concurrent(args):
url = f"{args.host}:{args.port}"
url = normalize_base_url(args.host, args.port)
prompt = "<image>\nPlease generate caption towards this image."
conv_template = copy.deepcopy(conv_llava_llama_3)
@@ -64,7 +66,7 @@ async def test_concurrent(args):
def test_streaming(args):
url = f"{args.host}:{args.port}"
url = normalize_base_url(args.host, args.port)
prompt = "<image>\nPlease generate caption towards this image."
conv_template = copy.deepcopy(conv_llava_llama_3)
conv_template.append_message(role=conv_template.roles[0], message=prompt)
@@ -104,7 +106,7 @@ def test_streaming(args):
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("--host", type=str, default="http://127.0.0.1")
parser.add_argument("--host", type=str, default="127.0.0.1")
parser.add_argument("--port", type=int, default=30000)
args = parser.parse_args()
asyncio.run(test_concurrent(args))

8
examples/runtime/multimodal/pixtral_server.py
View File

@@ -19,6 +19,8 @@ import json
import aiohttp
import requests
from sglang.utils import normalize_base_url
IMAGE_TOKEN_SEP = "\n[IMG]"
ROUTE = "/generate"
@@ -32,7 +34,7 @@ async def send_request(url, data, delay=0):
async def test_concurrent(args):
url = f"{args.host}:{args.port}{ROUTE}"
url = f"{normalize_base_url(args.host, args.port)}{ROUTE}"
# Single image test
if args.single_image:
@@ -69,7 +71,7 @@ async def test_concurrent(args):
def test_streaming(args):
url = f"{args.host}:{args.port}/generate"
url = f"{normalize_base_url(args.host, args.port)}/generate"
# Single image test
if args.single_image:
@@ -112,7 +114,7 @@ def test_streaming(args):
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("--host", type=str, default="http://127.0.0.1")
parser.add_argument("--host", type=str, default="127.0.0.1")
parser.add_argument("--port", type=int, default=30000)
parser.add_argument(
"--single-image",

8
examples/runtime/multimodal/qwen_llava_server.py
View File

@@ -21,6 +21,8 @@ import aiohttp
import requests
from llava.conversation import conv_qwen
from sglang.utils import normalize_base_url
async def send_request(url, data, delay=0):
await asyncio.sleep(delay)
@@ -31,7 +33,7 @@ async def send_request(url, data, delay=0):
async def test_concurrent(args):
url = f"{args.host}:{args.port}"
url = normalize_base_url(args.host, args.port)
prompt = "<image>\nPlease generate caption towards this image."
conv_template = copy.deepcopy(conv_qwen)
@@ -64,7 +66,7 @@ async def test_concurrent(args):
def test_streaming(args):
url = f"{args.host}:{args.port}"
url = normalize_base_url(args.host, args.port)
prompt = "<image>\nPlease generate caption towards this image."
conv_template = copy.deepcopy(conv_qwen)
conv_template.append_message(role=conv_template.roles[0], message=prompt)
@@ -104,7 +106,7 @@ def test_streaming(args):
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("--host", type=str, default="http://127.0.0.1")
parser.add_argument("--host", type=str, default="127.0.0.1")
parser.add_argument("--port", type=int, default=30000)
args = parser.parse_args()
asyncio.run(test_concurrent(args))