Arize provides machine learning observability and monitoring solutions for AI systems.

Arize

The AI observability & LLM Evaluation Platform.
 

Swift Ventures

Foundation Capital

Battery Ventures

The House Fund

Trinity Ventures

Arize AI

In tomorrow's paper read, Sally-Ann DeLucia and Julia Gomes will dive into a comprehensive study of LLM as a Judge. 👩‍⚖️👩‍⚖️ 🤖 

This can be a great tool for LLM evaluation, but there are still lots of open questions, so this paper takes a long look at the performance of LLMs acting as judges. 

The study included nine judge models and nine exam-taker models – both base and instruction-tuned. The judge models’ alignment is assessed across different model sizes, families, and judge prompts to answer questions about the strengths and weaknesses of this paradigm, and what potential biases it may hold.

Join us live: https://lnkd.in/dmEY6C8F

Want to ship with confidence, knowing your app is compliant and supplying relevant answers, while Also thwarting malicious inputs like jailbreak attempts? Tomorrow's webinar covers how to protect your LLM app. 🔒  

John Gilhuly and Safeer Mohiuddin (Co-Founder, Guardrails AI), will walk you through how you can use Guardrails and Arize to secure your app, while measuring the frequency and severity of any attacks and errors.

Register: https://lnkd.in/gn6ZpKHA

Given a collection of every kind of animal on earth - narwhals to puppies to mules, how do we find just the "pets"? How would we distinguish between a pack mule and a Moscow Mule? How is red as a color different from "red" as a term? And why is the famous example of (king-man+woman=queen) transferrable to examples like cities and countries, but flawed when recast as analogy?

Find out in this short talk by Jocelyn Matthews for Arize AI's Arize:Observe 2024 

Embeddings (Vectors) Explained: The Core Concepts for Developers - YouTube https://hubs.ly/Q02KtmNw0

Want to build and deploy multi-agent RAG? Gabriel P. presented on how Naologic did it at Observe last month. He went through prompt engineering, multi-agent architecture, agent-execution cycles, troubleshooting and more. 

Their team created agents for each job role: CEO, Head of Sales, Purchasing, Supply chain etc. To get high quality answers to business questions, they call each other and come up with an answer. 

Example: What is the best product to focus on to clear the inventory fast? CEO asks Sales, Sales needs some info from Supply Chain, gets the info, responds back and formats the final answer.

Watch here: https://lnkd.in/gXRgEUBi

How do you make it easy to approach AI applications and deploy them in the cloud? 

Kristen Womack, Principal Product Manager at Microsoft (on the Azure tools team) joined us at Observe this year to talk about how to build and deploy GenAI and LLM Ops apps to the cloud in a developer workflow. 

Womack also deploys Phoenix and shows how to visualize LLM evaluation and decision making. 🚀 

Find the full talk in the comments 👇

Building robust AI solutions requires more than just following a traditional workflow. 

Vibhu Sapra, AI Engineer at MindsDB, provided a brief overview of a typical AI development process and then dove into common misconceptions about what it takes to move from concept to production. 

Focusing on the often overlooked backend complexities, he revealed what truly goes into making an AI solution reliable and scalable in real-world applications.

Watch the full talk, originally given at Arize:Observe: https://lnkd.in/g_YYsatC

COM:ARIZE

Arize

People Not Available