EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

Bump the version to v0.1.8 (#93)

Browse Source
This commit is contained in:
Lianmin Zheng
2024-01-24 03:33:34 -08:00
committed by GitHub
parent 489796c7ea
commit c70b3cfa9e
4 changed files with 19 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
python/pyproject.toml
View File

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
[project] [project]
name = "sglang" name = "sglang"
version = "0.1.7" version = "0.1.8"
description = "A structured generation langauge for LLMs." description = "A structured generation langauge for LLMs."
readme = "README.md" readme = "README.md"
requires-python = ">=3.8" requires-python = ">=3.8"

2
python/sglang/__init__.py
View File

@@ -1,4 +1,4 @@
__version__ = "0.1.7" __version__ = "0.1.8"
from sglang.api import * from sglang.api import *
from sglang.global_config import global_config from sglang.global_config import global_config

18
python/sglang/srt/managers/tokenizer_manager.py
View File

@@ -50,7 +50,9 @@ def init_global_processor(server_args: ServerArgs):
) )
def get_pixel_values(image_data, image_aspect_ratio=None, image_grid_pinpoints=None, processor=None): def get_pixel_values(
image_data, image_aspect_ratio=None, image_grid_pinpoints=None, processor=None
):
try: try:
processor = processor or global_processor processor = processor or global_processor
image = load_image(image_data) image = load_image(image_data)
@@ -117,14 +119,22 @@ class TokenizerManager:
async def get_pixel_values(self, image_data): async def get_pixel_values(self, image_data):
aspect_ratio = getattr(self.hf_config, "image_aspect_ratio", None) aspect_ratio = getattr(self.hf_config, "image_aspect_ratio", None)
grid_pinpoints = self.hf_config.image_grid_pinpoints if aspect_ratio == "anyres" else None grid_pinpoints = (
self.hf_config.image_grid_pinpoints if aspect_ratio == "anyres" else None
)
if self.executor is not None: if self.executor is not None:
loop = asyncio.get_event_loop() loop = asyncio.get_event_loop()
return await loop.run_in_executor( return await loop.run_in_executor(
self.executor, get_pixel_values, image_data, aspect_ratio, grid_pinpoints self.executor,
get_pixel_values,
image_data,
aspect_ratio,
grid_pinpoints,
) )
else: else:
return get_pixel_values(image_data, aspect_ratio, grid_pinpoints, self.processor) return get_pixel_values(
image_data, aspect_ratio, grid_pinpoints, self.processor
)
async def generate_request(self, obj: GenerateReqInput): async def generate_request(self, obj: GenerateReqInput):
if self.to_create_loop: if self.to_create_loop:

5
python/sglang/srt/server.py
View File

@@ -380,12 +380,13 @@ def launch_server(server_args, pipe_finish_writer):
success = False success = False
for i in range(60): for i in range(60):
time.sleep(1)
try: try:
res = requests.get(url + "/get_model_info", timeout=5) res = requests.get(url + "/get_model_info", timeout=5)
success = True success = True
break break
except requests.exceptions.RequestException as e: except requests.exceptions.RequestException as e:
time.sleep(1) pass
if success: if success:
pipe_finish_writer.send("init ok") pipe_finish_writer.send("init ok")
@@ -406,7 +407,7 @@ class Runtime:
model_mode: List[str] = (), model_mode: List[str] = (),
schedule_heuristic: str = "lpm", schedule_heuristic: str = "lpm",
random_seed: int = 42, random_seed: int = 42,
log_level: str = "warning", log_level: str = "error",
): ):
host = "127.0.0.1" host = "127.0.0.1"
port = alloc_usable_network_port(1)[0] port = alloc_usable_network_port(1)[0]