TWIL: May 14, 2023
This week, I was catching up with the Azure Podcast with two new and interesting episodes. Also, an awesome article by Stephen Wolfram on how ChatGPT (and other LLMs) works, a video describing a state-of-the-art approach to improve GPT-4 results through Automatically Discovered Chain of Thought and a lot of news, both on the Azure front, and on the AI front as well. Namely, Google announced PaLM 2, Open AI announced Shap-E and Microsoft announced a host of new features for Bing and Edge around AI. Great times!
Episode 458: Integration Patterns
Elizabeth Graham, a Senior Software Engineer in the Commercial Software Engineering group at Microsoft, talks to Evan and Sujit about the various options we have for integrating external systems with Azure. She talks about how messages, files and other input forms can be ingested and processed with some of the services available in Azure.
Episode 459: AIOps
Evan and Sujit talk to Allison Jones, a Group PM for Azure’s AIOps System Team. She gives us insights into how Microsoft uses AI/ML to analyze various signals from Azure to provide actionable information about the state of the platform.
Large Language Models
What Is ChatGPT Doing … and Why Does It Work?
That ChatGPT can automatically generate something that reads even superficially like human-written text is remarkable, and unexpected. But how does it do it? And why does it work? My purpose here is to give a rough outline of what’s going on inside ChatGPT—and then to explore why it is that it can do so well in producing what we might consider to be meaningful text.
GPT 4 is Smarter than You Think: Introducing SmartGPT
In this video, I will not only show you how to get smarter results from GPT 4 yourself, I will also showcase SmartGPT, a system which I believe, with evidence, might help beat MMLU state of the art benchmarks. This should serve as your ultimate guide for boosting the automatic technical performance of GPT 4, without even needing few shot exemplars. The video will cover papers published in the last 72 hours, like Automatically Discovered Chain of Thought, which beats even ‘Let’s think Step by Step’ and the approach that combines it all.
Azure Data Explorer for Vector Similarity Search
In the world of AI & data analytics, vector databases are emerging as a powerful tool for managing complex and high-dimensional data. In this article, we will explore the concept of vector databases, the need for vector databases in data analytics, and how Azure Data Explorer (ADX) aka Kusto can be used as a vector database.
Public preview: Azure Cold Storage
Azure Blob Storage is optimized for storing massive amounts of unstructured data. With blob access tiers, you can store your blob data in the most cost-effective manner based on how frequently it will be accessed and how long it will be retained. Now Azure Blob Storage adds a new online access tier, cold, in addition to hot, cool and archive.
Generally available: Serverless SQL for Azure Databricks
Serverless SQL for Azure Databricks is now generally available and will be enabled for your eligible workspaces between now and the end of May. This feature provides instant elastic compute to users for their BI and SQL workloads, with minimal management required and capacity optimizations that can lower overall costs. You only pay for Serverless SQL when users start running reports or queries. A Serverless SQL warehouse includes compute cost. Serverless SQL is built for modern business analysts who work on their own schedules and want instant compute available to process queries without waiting for clusters to start up or scale out.
GPT – ALE (Generative Pre-trained Transformer – Active Learning Engine)
Data labeling and categorization is a critical step in the machine learning lifecycle. However, it is often a time-consuming and expensive process. GPT-ALE reduces the time and cost of data labeling, by combining state-of-the-art foundation models (e.g. GPT), active learning, and regularized self-training. It uses active learning strategies to identify samples that help it simultaneously understand the user’s intent and the dataset, and offers suggestions to the user as to how to label the data.
Data Discovery Solution
The Data Discovery Playbook aims to quickly provide structured views on your text, images and videos, all at scale using Synapse and unsupervised ML techniques that exploit state of the art deep learning models. The goal is to present this data to and facilitate discussion with a business user/data owner very quickly via PowerBI visualisation, so that the customer and team can decide the next best action with the data, identify outliers or generate a training data set for a supervised model.
OpenAI’s new Shap-E tool is Dall-E for 3D objects
OpenAI‘s latest endeavor, Shap-E, is a model that allows you to generate 3D objects from text, not unlike how Dall-E can create 2D images. According to OpenAI, Shap-E is “a conditional generative model for 3D assets. Unlike recent work on 3D generative models which produce a single output representation, Shap-E directly generates the parameters of implicit functions that can be rendered as both textured meshes and neural radiance fields.”
Introducing PaLM 2
PaLM 2 is more capable, it’s also faster and more efficient than previous models — and it comes in a variety of sizes, which makes it easy to deploy for a wide range of use cases. We’ll be making PaLM 2 available in four sizes from smallest to largest: Gecko, Otter, Bison and Unicorn. Gecko is so lightweight that it can work on mobile devices and is fast enough for great interactive applications on-device, even when offline. This versatility means PaLM 2 can be fine-tuned to support entire classes of products in more ways, to help more people.
Announcing the next wave of AI innovation with Microsoft Bing and Edge
Today I’m thrilled to share we are moving to the next generation of AI-powered Bing and Edge to transform the largest category of software in the world – search – by greatly expanding the vision and capabilities we think of as Your Copilot for the Web.
Prompt Engineering Logo
This is an emerging field, and there are more and more resources available such as courses, presentations, and articles. Now is the perfect time to start learning and utilizing it in your projects. Additionally, now is an excellent time to have a logo for it! I created the logo based on my vision and dedicated it to the public domain, so anyone can legally use it without permission.
Humane’s new wearable AI demo is wild to watch — and we have lots of questions
Buzz has been building around the secretive tech startup Humane for over a year, and now the company is finally offering a look at what it’s been building. At TED last month, Humane co-founder Imran Chaudhri gave a demonstration of the AI-powered wearable the company is building as a replacement for smartphones. Bits of the video leaked online after the event, but the full video is now available to watch.
Have a brilliant week!