Skip to main content

轻量级 LLM 应用可观测性 SDK — 调用追踪、提示词工程、批量处理

Project description

UltraFlow

轻量级 LLM 应用可观测性 SDK — 调用追踪、提示词工程、批量处理。

特性

  • Prompty — 将 .prompty 文件作为无状态函数执行(同步 / 异步 / 流式)
  • Tracing@trace 装饰器自动采集调用树,业务代码零侵入
  • 批量处理FlowProcessor 并发执行批量任务
  • CLIuf inituf newuf run,快速上手

安装

pip install UltraFlow

快速开始

1. 初始化项目

uf init my-project
cd my-project

2. 执行 Prompty

from ultraflow import Prompty

translate = Prompty("translate.prompty")
result = translate(text="Hello, world!")

3. 添加调用追踪

from ultraflow import Prompty, configure_tracing, shutdown_tracing, trace

configure_tracing(endpoint="http://localhost:8031")

@trace(kind="flow")
def translate_flow(text: str) -> str:
    prompty = Prompty("translate.prompty")
    return prompty(text=text)

result = translate_flow(text="Hello!")
shutdown_tracing()

4. 批量处理

from ultraflow import Prompty, FlowProcessor

translate = Prompty("translate.prompty")
processor = FlowProcessor(translate, "data.json", max_workers=5)
results = processor.run()

CLI 命令

命令 说明
uf init [PROJECT] 初始化项目,生成配置和示例 prompty
uf new FLOW_NAME 创建 .prompty + .json 模板
uf run FLOW_NAME 运行流程(自动查找同名 .json
uf run FLOW -d DATA -w N 指定数据文件和并发数
uf connection init 生成连接配置模板
uf connection list 显示当前生效的连接配置

License

MIT

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

ultraflow-1.0.7.tar.gz (25.9 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

ultraflow-1.0.7-py3-none-any.whl (23.2 kB view details)

Uploaded Python 3

File details

Details for the file ultraflow-1.0.7.tar.gz.

File metadata

  • Download URL: ultraflow-1.0.7.tar.gz
  • Upload date:
  • Size: 25.9 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: pdm/2.26.7 CPython/3.12.11 Darwin/24.6.0

File hashes

Hashes for ultraflow-1.0.7.tar.gz
Algorithm Hash digest
SHA256 70e9438e307a7f936625aedd302e12042ebef6a00f9ff9b6eba2b1ca6f1b2d4b
MD5 cb955c957c4b5bcdc46df2d9538ec082
BLAKE2b-256 952cf003690fabebe116f86ca79b285840c8e3c1f11215b051f006e2ca8d0dcd

See more details on using hashes here.

File details

Details for the file ultraflow-1.0.7-py3-none-any.whl.

File metadata

  • Download URL: ultraflow-1.0.7-py3-none-any.whl
  • Upload date:
  • Size: 23.2 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: pdm/2.26.7 CPython/3.12.11 Darwin/24.6.0

File hashes

Hashes for ultraflow-1.0.7-py3-none-any.whl
Algorithm Hash digest
SHA256 e60e1f49291c63bf444339bae963dd95e7139ef1b12342b50abc31e65a8c7842
MD5 7e091245e4f5d152c42c2b46733232cc
BLAKE2b-256 b5a11083787a3850bd0a02155288742645fe582ac7a51cf3ce1b4697909a4852

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page