The OpenAI Assistant Starter Kit is an open-source starter kit that you can use to kickstart your development of OpenAI chat applications.

The post OpenAI Assistant Starter Kit Updated with AssistantStream appeared first on OpenAI Blog - Stephen Walther on OpenAI.

Alongside his day job as The MagPi and Hackspace magazines’ publisher, Brian Jepson likes to track down and restore old computers. As a volunteer at Rhode Island’s Computer Museum, he regularly rescues old computers “from 8-bit on up”. A favourite is the Atari ST line of 16/32-bit computers which launched in 1985, featured Motorola’s 68000 CPU range and, “uniquely, could be used with electronic synthesisers as well as in home computers”. Brian’s mt32pi synth project shows just how well the 40-year-old hardware works with a thoroughly modern Raspberry Pi. 

Musical inspiration

Brian used to enjoy playing music (and regrets not having kept it up) and was keen to add a vintage synthesiser to his collection. He observes that the Atari ST’s MIDI ports were notable because certain videogames from the era were capable of producing high-quality background music with a MIDI synthesiser. “There were many music sequencers you could use to compose music with a MIDI keyboard and play it on a MIDI synthesiser. However, the Atari ST was unique in that it had built-in MIDI ports for connecting to electronic musical instruments as well as synthesisers.” There are also several MIDI-capable keyboards that were designed as videogame controllers that, for some reason, include the five-pin MIDI DIN cable, Brian continues. “For example, the Rock Band 3 Wireless Keyboard for Nintendo Wii is a relatively affordable keyboard that works great with this setup! It’s unusual to see the five-pin connector because a lot of modern MIDI stuff uses USB now.”

For his own needs, Brian “looked into buying a used Roland MT32 synthesiser, which was roughly contemporary with the Atari ST, but decided the versatility of a Raspberry Pi-based solution was preferable”. For his Atari ST revival, Brian used the ClumsyMIDI board with mt32-pi software – an MT-32 emulator that works with Raspberry Pi 3, 4 and Zero. Developer Dale’s wiki explains the project in detail: it was ready to go and worked well with Brian’s Raspberry Pi 3Bs and “a couple of Ataris I’ve outfitted with mt32-pi synthesisers”. 

Fast and affordable

Dale Whinham’s open source mt32-pi software is a bare-metal MIDI synthesiser that operates as a kernel. It boots directly into the mt32-pi software, starts up in seconds, doesn’t mind being unplugged without the usual shutdown process, is extremely fast, and is far more versatile than any alternative Brian could find.

Next, Brian needed to decide which board would work best for the Atari ST rebuild. He chose ClumsyMidi because it is based entirely on through-hole parts which he was able to order along with the Clumsy circuit boards.

Aside from the Atari ST, Brian estimates the project cost roughly $75 and came together very easily. There was one aspect that left him a little dissatisfied though. “If you’re planning to mess around with making music on your Atari, you will have the most fun if you can compose using a real musical instrument rather than just using the software. So you’ll want to build some practice and learning into your time budget. I say this as someone who has neglected that part and regrets it.” 

Brian’s Atari ST is used for both retro games and MIDI demos at the Rhode Island Computer Museum, where visitors can connect it to a keyboard and try out different instruments playing the freeware Sweet Sixteen sequencer. “It’s amazing how much fun someone can have switching instruments and banging on keys.” 

The MagPi #141 out NOW!

You can grab the brand-new issue right now from Tesco, Sainsbury’s, Asda, WHSmith, and other newsagents, including the Raspberry Pi Stores in Cambridge. It’s also available at our online store which ships around the world. You can also get it via our app on Android or iOS.

You can also subscribe to the print version of The MagPi. Not only do we deliver it globally, but people who sign up to the six- or twelve-month print subscription get a FREE Raspberry Pi Pico W!

The post Mt32-pi Atari ST | The MagPi #141 appeared first on Raspberry Pi.

Introduction:
As a technical student, you're likely fascinated by the rapidly evolving world of artificial intelligence (AI) and machine learning. In this blog, we'll explore the exciting realm of Small Language Models (SLMs), a new breed of AI models that are redefining what's possible with Azure.

What are Small Language Models (SLMs)?
Small Language Models (SLMs) are a type of neural network designed to process and generate human-like text. Unlike traditional language models, SLMS are optimized for smaller size and faster training times, making them more accessible and scalable for developers like you.

Introducing Phi-3: Redefining What's Possible with SLMs
Microsoft announced the introduction of Phi-3, a revolutionary new architecture that leverages the power of SLMS. Phi-3 redefines what's possible with language models, enabling faster and more accurate text processing.

Getting Started with Phi-3 and Azure AI Studio: A Step-by-Step Guide

Step 1: Set Up Your Environment

Create an Azure account if you haven't already. 
Install the Azure CLI or use a cloud-based IDE like Visual Studio Code.
Familiarize yourself with Python, as it's the primary language used in Azure AI Studio.

Step 2: Setup your Azure Azure AI Studio

Go to the Azure AI Studio and follow the installation instructions for your preferred platform (Windows, macOS, or Linux).
Launch the Azure AI Studio and sign in with your Azure account credentials.

Step 3: Explore Phi-3 Models

Explore the model catalog in Azure AI Studio - Azure AI Studio
Navigate to the Model Catalog and install the Phi-3 models.
Import the Phi-3 models into Azure AI Studio using the built-in data loading features.
Experiment with different Phi-3 model variants to see how they perform on various text processing tasks.

Step 4: Build Your First SLM-based Application

Choose a sample project from the Azure AI Studio template gallery or create your own from scratch.
Use the Phi-3 models and Azure AI Studio's built-in functionality to develop an SLM-based application that performs text classification, sentiment analysis, or language translation.

Step 5: Deploy Your Application

Once you've developed your SLM-based application, deploy it to the cloud using Azure Kubernetes Service (AKS) or Azure Functions.
Use Azure Monitor and Log Analytics to track your application's performance and troubleshoot any issues that arise.

Tips and Tricks for Success

Experiment with different Phi-3 model variants and hyperparameters to optimize your SLM-based application's performance.
Leverage Azure AI Studio's built-in collaboration features, such as code sharing and version control, to work with team members on your project.
Join the Microsoft AI Discord community forum to connect with other developers and get help when you need it.

Conclusion:
In this blog, we've explored the exciting world of Small Language Models (SLMs) and introduced you to Phi-3, a revolutionary new architecture that leverages the power of SLMS. By following the steps outlined in this guide, you can get started with Phi-3 and Azure AI Studio and unlock the full potential of SLMs for your technical projects.

Additional Resources:
Try Phi-3 Playground Experience
Azure Blog: Introducing Phi-3: Redefining what's possible with SLMs | Microsoft Azure Blog
Microsoft Developer Blog: Phi-3 Models Getting Started
Azure AI Studio Documentation: Getting Started Guide
Phi-3  benchmarks and technical paper. 
Phi-3 on the AI Show

Microsoft AI Discord Community http://aka.ms/AzureAI/Discord

Global AI Community http://globalai.community

Azure AI Samples http://github.com/azure-samples/azureai-samples

Microsoft AI Show http://aka.ms/aishow

Paul Jarvis: Key Indicators of Success for Agile Teams and Their Scrum Masters That You Can Reflect On

Read the full Show Notes and search through the world’s largest audio library on Scrum directly on the Scrum Master Toolbox Podcast website: http://bit.ly/SMTP_ShowNotes.

Investigate Paul's perspective on what defines success for Scrum Masters, from the smooth running of refinement sessions to the collaborative maintenance of the Product Backlog. This episode provides practical tips for Scrum Masters to assess their impact, emphasizing the importance of team collaboration, effective backlog management, and the ease of sprint planning. Discover how to assess success in the Agile world through a blend of team dynamics, process efficiency, and shared responsibility.

Featured Retrospective Format for the Week: The Sailboat Retrospective and Other Formats

Paul emphasizes the critical role of retrospectives in enabling relentless improvement, highlighting formats like the Sailboat retrospective, Celebration Grid, and the high-performance tree to encourage experimentation and self-assessment within teams. These retrospective formats are not just reflective exercises but strategic tools that prompt teams to align their values with their Agile practices, ensuring continuous growth and alignment with business goals.

[IMAGE HERE] Retrospectives, planning sessions, vision workshops, we are continuously helping teams learn about how to collaborate in practice! In this Actionable Agile Tools book, Jeff Campbell shares some of the tools he’s learned over a decade of coaching Agile Teams. The pragmatic coaching book you need, right now! Buy Actionable Agile Tools on Amazon, or directly from the author, and supercharge your facilitation toolbox! 

About Paul Jarvis

Paul is a seasoned Enterprise Lean Agile Coach, Trainer, RTE, and Scrum Master with a decade of experience in the FinTech sector, focusing on banking, payments, and e-commerce. Recently, he completed a 3.5-year tenure at a key player in investment banking.

You can link with Paul Jarvis on LinkedIn and connect with Paul Jarvis on Twitter.

Host(s):

• John Papa @John_Papa
• Craig Shoemaker @craigshoemaker

Guest:

• Alfredo Deza on LinkedIn

Recording date: Apr 11, 2024

Brought to you by

• AG Grid

  

• IdeaBlade

Resources:

• Alfredo’s social media profile at LinkedIn
• Alfredo Deza at the Olympics
• Alfredo Deza’s Olympian story
• Alfredo’s father in the Olympics
• Alfredo Deza books on Amazon
• Alfredo’s Books at O’Reilly
• Alfredo Deza at Duke University
• Getting Started with Testing in Python
• Python Fundamentals on Khan Academy
• Kubernetes
• Small functions considered harmful
• Reddit thread on small functions being harmful
• Writing small functions
• Five lines of code principle
• Python clean code - 6 best practices to make your Python functions more readable
• Refactoring to Patterns

Timejumps

• 00:28 Welcome!
• 01:35 Introducing Alfredo Deza
• 02:59 What's your Olympic story?
• 06:02 What lessons from being an athlete into technology?
• 09:51 Sponsor: Ag Grid
• 10:48 Why Python?
• 23:29 Software engineers dislike writing tests
• 26:13 How should developers balance the principles of testing?
• 31:34 Happy birthday to Ward
• 32:44 How do you break functions into multiple functions?
• 36:46 Sponsor: IdeaBlade
• 37:36 Where should people go to learn more?

Podcast editing on this episode done by Chris Enns of Lemon Productions.