12.07.2023

What's New

New AWS Bedrock Integration for Anthropic Models

Users can now iterate on prompts in the Prompt Playground using the AWS Bedrock integration. This integration allows users to iterate on prompt templates, parameters, and variables in the platform and compare responses. Additionally, users can now compare LLM providers by comparing prompt runs between LLMs. Learn more here.

ROC Curve

Users can now visualize an ROC curve with calculated AUC for their model from within Arize on the Performance Tracing page. Simply select the "More Charts" tab, and then "ROC Curve".

Feature Drift Dashboard Template

A new dashboard template has been added for easier analysis of drift across all your model features. To access this new template, create a new dashboard, select the "Drift" tab, and then select the "Feature Drift" template.

List of Strings

  • Able to ingest a list of strings ([item_1, item_2, item_3]) and view metrics

  • New metrics support: Percent Empty and Average List Length (data quality metric)

  • Filters: on every filter page you can select contains

  • Valid for all ingestion methods (except CSV)

Enhancements

New Side Navigation

Model navigation can now be found (and collapsed) on the left side of the UI for an improved experience. With this design, we have also categorized groupings of tabs so you can more easily find what you're looking for.

New Greyscale Dark Mode

Arize's new greyscale dark mode new dark mode introduces a richer, greyscale palette that minimizes distractions and is optimized for work in low-light environments, reducing eye strain for extended evening hours.

View Dashboard Line Plot by Feature or Tag

To understand model performance across features/tags of your data, users now can group by features or tags on dashboard line plots. To do this, simply create a Time Series plot, and turn on "Group metric by feature or tag".

Model Loading Page

To help users better understand where there data is during data ingestion, users will now be navigated to a loading page as Arize ingests their data.

🎓 New Educational Content

The latest ebooks, self-guided course modules, and technical posts on topics like LLM evaluation and beyond:

Last updated