Prompt Chain

Instrument LLM Apps with Phoenix Tracing

Arize-Phoenix LlamaPack is a multi-step prompt workflow integration for LlamaIndex that connects query engines with the Arize Phoenix observability platform.

Works with openaiarize

91
Spark score
out of 100
Updated last month
Version 0.14.22

Add to Favorites

Why it matters

Instrument your LlamaIndex applications for comprehensive LLM tracing and observability using the open-source Phoenix library. Gain insights into your LLM's performance and behavior.

Outcomes

What it gets done

01

Integrate Phoenix for LLM tracing within LlamaIndex applications.

02

Download and utilize the ArizePhoenixQueryEnginePack via CLI or code.

03

Parse documents and run queries to generate trace data.

04

Access Phoenix session URLs to view detailed trace information.

Install

Add it to your toolbox

Run in your project directory:

curl -fsSL https://spark.entire.vc/get/li-pack-packs-arize-phoenix-query-engine | bash

Steps

Steps in the chain

01
Download the LlamaPack
02
Configure OpenAI API Key
03
Parse Documents into Nodes
04
Initialize the LlamaPack
05
Run Queries
06
View Trace Data in Phoenix UI
07
Access LlamaPack Internals

Overview

Arize-Phoenix LlamaPack

The Arize-Phoenix LlamaPack is a workflow integration that connects LlamaIndex query engines with the Arize Phoenix observability platform. It instruments query pipelines to enable tracing, performance monitoring, and debugging of retrieval-augmented generation systems. Use this pack when building production LlamaIndex applications that need observability for query engines, particularly when tracing multi-step retrieval workflows or evaluating RAG pipeline performance. It is most valuable for teams debugging complex query execution or analyzing the quality of retrieved context.

What it does

The Arize-Phoenix LlamaPack provides a workflow integration between LlamaIndex query engines and Arize Phoenix, an observability and evaluation platform for LLM applications. This pack enables developers to instrument their LlamaIndex query pipelines with Phoenix's tracing and monitoring capabilities, allowing them to track query execution, analyze performance, and debug issues in their retrieval-augmented generation (RAG) systems.

When to use - and when NOT to

Use this pack when you are building production LlamaIndex applications that require observability, performance monitoring, or debugging capabilities for your query engines. It is particularly valuable when you need to trace multi-step retrieval workflows, identify bottlenecks in your RAG pipeline, or evaluate the quality of retrieved context and generated responses. Do not use this pack if you are building simple proof-of-concept applications where observability overhead is unnecessary, or if you are already using a different observability solution that conflicts with Arize Phoenix's instrumentation approach.

Source README

Description pending for li-pack-packs-arize-phoenix-query-engine.

FAQ

Common questions

Discussion

Questions & comments · 0

Sign In Sign in to leave a comment.