Stability AI Reveals 'StableLM Zephyr 3B', 60% Smaller Yet Accurate
Plus: Meta launches Purple Llama for Safe AI, Meta's new update to Codec Avatars.
Hello Engineering Leaders and AI Enthusiasts!
Welcome to the 164th edition of The AI Edge newsletter. This edition brings you Stability AI’s new StableLM Zephyr 3B, Which is 60% smaller yet accurate.
And a huge shoutout to our incredible readers. We appreciate you😊
In today’s edition:
🌟 Stability AI reveals StableLM Zephyr 3B, 60% smaller yet accurate
🦙Meta launches Purple Llama for Safe AI development
👤 Meta released an update to Codec Avatars with lifelike animated faces
📚 Knowledge Nugget: Welcome to the World of Small(er) Language Models by
Let’s go!
We need your help!
We are working on a Gen AI survey and would love your input.
It takes just 2 minutes.
The survey insights will help us both.
And hey, you might also win a $100 Amazon gift card!
Every response counts. Thanks in advance!
Stability AI reveals StableLM Zephyr 3B, 60% smaller yet accurate
StableLM Zephyr 3B is a new addition to StableLM, a series of lightweight Large Language Models (LLMs). It is a 3 billion parameter model that is 60% smaller than 7B models, making it suitable for edge devices without high-end hardware. The model has been trained on various instruction datasets and optimized using the Direct Preference Optimization (DPO) algorithm.
It generates contextually relevant and accurate text well, surpassing larger models in similar use cases. StableLM Zephyr 3B can be used for a wide range of linguistic tasks, from Q&A-type tasks to content personalization, while maintaining its efficiency.
Why does this matter?
Tested on platforms like MT Bench and AlpacaEval, StableLM Zephyr 3B shows it can create text that makes sense, fits the context, and is linguistically accurate. In these tests, it competes well with bigger models like Falcon-4b-Instruct, WizardLM-13B-v1, Llama-2-70b-chat, and Claude-V1.
Meta launches Purple Llama for Safe AI development
Meta has announced the launch of Purple Llama, an umbrella project aimed at promoting the safe and responsible development of AI models. Purple Llama will provide tools and evaluations for cybersecurity and input/output safeguards. The project aims to address risks associated with generative AI models by taking a collaborative approach known as purple teaming, which combines offensive (red team) and defensive (blue team) strategies.
The cybersecurity tools will help reduce the frequency of insecure code suggestions and make it harder for AI models to generate malicious code. The input/output safeguards include an openly available foundational model called Llama Guard to filter potentially risky outputs.
This model has been trained on a mix of publicly available datasets to enable the detection of common types of potentially risky or violating content that may be relevant to a number of developer use cases. Meta is working with numerous partners to create an open ecosystem for responsible AI development.
Why does this matter?
Meta’s strategic shift toward AI underscores its commitment to ethical AI. Their collaborative approach to building a responsible AI environment emphasizes the importance of enhancing AI safety, which is crucial in today's rapidly evolving tech landscape.
Meta released an update to Codec Avatars with lifelike animated faces
Meta Research’s work presents Relightable Gaussian Codec Avatars, a method to create high-quality animated head avatars with realistic lighting and expressions. The avatars capture fine details like hair strands and pores using a 3D Gaussian geometry model. A novel relightable appearance model allows for real-time relighting with all-frequency reflections.
The avatars also have improved eye reflections and explicit gaze control. The method outperforms existing approaches without sacrificing real-time performance. The avatars can be rendered in real-time from any viewpoint in VR and support interactive point light control and relighting in natural illumination.
Why does this matter?
With the help of Codec Avatars soon, this technology will enable us to communicate with someone as if they were sitting across from us, even if they're miles apart. Also, This leads to incredibly detailed real-time avatars, precise down to individual hair strands!
Enjoying the daily updates?
Refer your pals to subscribe to our daily newsletter and get exclusive access to 400+ game-changing AI tools.
When you use the referral link above or the “Share” button on any post, you'll get the credit for any new subscribers. All you need to do is send the link via text or email or share it on social media with friends.
Knowledge Nugget: Welcome to the World of Small(er) Language Models
In this article, the author
explains that the emergence of small(er) language models challenges the idea that bigger is always better in generative AI. SLMs are smaller, highly specialized, and more affordable models that are gaining traction in specific use cases.Microsoft and Meta are leading the SLM movement, with Microsoft releasing Phi-2 for mathematical reasoning and Orca2 for reasoning tasks. As large language models (LLMs) continue to scale, SLMs offer more control, optimization, and cost-effectiveness. In other news, there are updates on ML research, AI tech releases, real-world ML applications, and recent developments in the AI industry.
Why does this matter?
Smaller, specialized models like SLMs from Microsoft and Meta offer precision and cost-effectiveness in specific tasks, signaling a shift in AI development towards more controlled and efficient models. The piece also covers updates in AI research, tech releases, and real-world applications, highlighting ongoing advancements.
What Else Is Happening❗
🤑 AMD predicts the market for its data center AI processors will reach $45B
An increase from its previous estimate of $30B, the company also announced the launch of 2 new AI data center chips from its MI300 lineup, one for generative AI applications and another for supercomputers. AMD expects to generate $2B in sales from these chips by 2024. (Link)
📱 Inflection AI’s Pi is now available on Android!
The Android app is available in 35 countries and offers text and hands-free calling features. Pi can be accessed through WhatsApp, Facebook Messenger, Instagram DM, and Telegram. The app also introduces new features like back-and-forth conversations and the ability to choose from 6 different voices. (Link)
🚀 X started rolling Grok to X premium users in the US
Grok uses a generative model called Grok-1, trained on web data and feedback from human assistants. It can also incorporate real-time data from X posts, giving it an advantage over other chatbots in providing up-to-date information. (Link)
🎨 Google Chrome could soon let you use AI to create a personalized theme
The latest version of Google Chrome Canary includes a new option called 'Create a theme with AI’, which replaces the 'Wallpaper search' option. An 'Expanded theme gallery' option will also be available, offering advanced wallpaper search options. (Link)
🖼️ Pimento uses AI to turn creative briefs into visual mood boards
French startup Pimento has raised $3.2M for its gen AI tool that helps creative teams with ideation, brainstorming, and moodboarding. The tool allows users to compile a reference document with images, text, and colors that will inspire and guide their projects. (Link)
That's all for now!
If you are new to The AI Edge newsletter, subscribe to get daily AI updates and news directly sent to your inbox for free!
Thanks for reading, and see you tomorrow. 😊