lixiaolong
b386f003a0
- 集成Qwen3.8-27B-FP8快速模型和Qwen3.8-27B精确模型作为SGLang服务 - 添加SGLang API配置选项(SGLANG_FAST_BASE_URL、SGLANG_ACCURATE_BASE_URL等) - 实现音频分类中的决策聚合算法(topK、nearestWeight、labelMaxDistance) - 添加视频采样帧限制(VIDEO_SAMPLING_FRAME_LIMIT)和上下文token限制 - 更新健康检查以监控SGLang服务状态 - 实现视频分析的双模式决策策略(快速+精确) - 添加音频参考文件导入工具(import_audio_references.py) - 扩展音频分类标签支持FIND_VEHICLE_HORN类别 - 优化视频分析的帧采样策略，始终包含视频尾部帧 - 添加决策策略参数(tuning、decisionPolicy)支持 - 更新配置类以支持新的SGLang和视频参数 - 修改compose配置以支持Qwen3.8模型部署 - 更新音频分类测试用例验证聚合逻辑 - 重构视频测试以支持SGLang API格式和决策策略
29 lines
923 B
Python
29 lines
923 B
Python
from pathlib import Path
|
|
|
|
from pydantic_settings import BaseSettings, SettingsConfigDict
|
|
|
|
|
|
class Settings(BaseSettings):
|
|
model_config = SettingsConfigDict(env_file=".env", extra="ignore")
|
|
|
|
api_key: str = ""
|
|
profiles_dir: Path = Path("/data/profiles")
|
|
artifacts_dir: Path = Path("/data/artifacts")
|
|
jobs_dir: Path = Path("/data/jobs")
|
|
ollama_base_url: str = "http://host.docker.internal:11434"
|
|
ollama_model: str = "qwen3-vl:32b"
|
|
media_timeout_seconds: int = 120
|
|
ollama_timeout_seconds: int = 600
|
|
sglang_fast_base_url: str = ""
|
|
sglang_accurate_base_url: str = ""
|
|
sglang_api_key: str = ""
|
|
sglang_timeout_seconds: int = 600
|
|
vision_ollama_fallback_enabled: bool = True
|
|
max_audio_bytes: int = 100 * 1024 * 1024
|
|
max_video_bytes: int = 2 * 1024 * 1024 * 1024
|
|
video_sampling_frame_limit: int = 100
|
|
max_video_context_tokens: int = 262144
|
|
|
|
|
|
settings = Settings()
|