04.17.2023
New Releases, Enhancements, Changes + Arize in the News!
What's New
Cluster Download for Active Learning
Once a cluster that is impacting model performance has been identified, download the data in the cluster for active learning. This data includes all the information needed for labeling workflows. These clusters are highly focused groups of data points, enabling labeling teams to be more precise in their efforts.
UMAP Filters for Additional Insights
Gain a better visual understanding of your data, and discover new insights with UMAP filters. Use filters to focus in on a specific slice of your data in prior to generating the UMAP.
Enhancements
Cloud Storage & Data Warehouse Ingestion
UX Updates: Enhanced UX to easily parse file path and copy access permissions. Simply paste your file path, and Arize will parse your bucket name and prefix.
File Preview: Once you've configured your bucket access permissions, preview your file to validate expected column names and data match the 'Schema Mapping' fields.
Monitors UI: Edit & Custom Monitors
Revised 'Monitors' UI to break down monitor configurations in a few easy steps. Use the editable fields to configure filters, a delayed evaluation window, alerting threshold, notifications, and more.
In The News
New Course Updates!
Arize's self-guided ML observability course continues to add modules and practical content on the emerging field of LLMOps along with other core topics.
New additions focused on generative AI include:
🦜 LangChain: how to fine-tune large language model (LLM) applications and maximize LLM production performance with LangChain
🔵 Bleu Score and LLM Metrics: tips for measuring text-based generative models using BLEU, ROUGE, METEOR, and Bertscore as well as prediction embeddings
📝 A Novel Approach for Applying LLMs To Tabular Data: an adventure in unleashing large language models (LLMs) on tabular Kaggle competitions
➡️ Transfer Learning: a primer on a pivotal concept
Other additions include:
Data Binning: How to bin to win and overcome feature binning challenges
Collaborative Filtering: how to monitor your recommendation systems
F1 Score: when and where to use this important metric
For the third consecutive year, Arize AI made the Forbes AI 50 list. This year's list includes a special focus on companies redefining what's possible through generative AI.
Guide: Shifting To Automated, Dynamic Model Retraining
Today, many models are in production with no retraining at all, use manual retraining methods, or are retraining without optimizing or studying the cadence. This guide can help you change the game.
In this third episode of Deep Papers, a podcast from AI Pub and Arize AI, we interview Timo Schick and Thomas Scialom, the Research Scientists at Meta AI behind Toolformer.
Last updated
