Meta Unveils Llama 2, A Worthy Rival to ChatGPT
Plus: Microsoft's updates to further AI ambitions, How is ChatGPT's behavior changing over time.
Hello Engineering Leaders and AI Enthusiasts!
Welcome to the 66th edition of The AI Edge newsletter. This edition brings you Meta’s Llama 2, a free model that may be a worthy competitor for OpenAI’s ChatGPT.
Our amazing readers deserve a special mention. Your support is invaluable to us!😊
In today’s edition:
🚀 Meta unveils Llama 2, a worthy rival to ChatGPT
🌟 Microsoft furthers its AI ambitions with major updates
💬 How is ChatGPT's behavior changing over time?
📚 Knowledge Nugget: Reinforcement Learning from Human Feedback (RLHF) and LLMs: The Magic Sauce behind ChatGPT by
Let’s go!
Meta unveils Llama 2, a worthy rival to ChatGPT
Meta has introduced Llama 2, the next generation of its open-source large language model. Here’s all you need to know:
It is free for research and commercial use. You can download the model here.
Microsoft is the preferred partner for Llama 2. It is also available through AWS, Hugging Face, and other providers.
Llama 2 models outperform open-source chat models on most benchmarks tested, and based on human evaluations for helpfulness and safety, they may be a suitable substitute for closed-source models.
Meta is opening access to Llama 2 with the support of a broad set of companies and people across tech, academia, and policy who also believe in an open innovation approach for AI.
Why does this matter?
Meta had been making headlines about launching a freely available AI and this brings that vision to life. It also seems Llama 2 is the first fully open-source, real competitor for OpenAI’s ChatGPT.
However, its significance lies in its accessibility. It will democratize AI technology, level the playing field for AI development, and foster widespread innovation.Microsoft furthers its AI ambitions with major updates
At Microsoft Inspire, Meta and Microsoft announced support for the Llama 2 family of LLMs on Azure and Windows. In other news, Microsoft announced major updates for AI-powered Bing, Copilot, and more.
It announced Bing Chat Enterprise, which gives organizations AI-powered chat for work with commercial data protection.
Microsoft 365 Copilot will now be available for commercial customers for $30 per user per month. Copilot is also coming to Teams phone and chat.
It launched Vector Search in preview through Azure Cognitive search, which will capture the meaning and context of unstructured data to make search faster.
It is rolling out multimodal capabilities via Visual Search in Chat. Leveraging OpenAI’s GPT-4 model, the feature lets anyone upload images and search the web for related content.
Why does this matter?
Microsoft's extended partnership with Meta could establish it as the world’s supercomputing platform for AI. Plus, these groundbreaking announcements are proof of Microsoft’s commitment to integrating the best AI models, features, and tools into its business suite.How is ChatGPT's behavior changing over time?
GPT-3.5 and GPT-4 are the two most widely used LLM services, but how updates in each affect their behavior is unclear. A new study evaluated the behavior of the March 2023 and June 2023 versions of GPT-3.5 and GPT-4 on four tasks. And here are the findings:
Solving math problems- GPT-4 got much worse, while GPT-3.5 greatly improved.
Answering sensitive/dangerous questions- GPT-4 became less willing to respond directly, while GPT-3.5 was slightly more willing.
Code generation- Both systems made more mistakes that stopped the code from running in June compared to March.
Visual reasoning- Both systems improved slightly from March to June.
It shows that the behavior of the same LLM service can change substantially in a relatively short period (and for the worse in some tasks), highlighting the need for continuous monitoring of LLM quality.
Why does this matter?
The study attempts to answer whether LLM services like GPT4 consistently do get “better” over time and whether updates to the models aimed at improving some aspects actually hurt their capability in other dimensions. It encourages companies who rely on LLM services as a component in their ongoing workflow to implement similar monitoring analysis for better results.Knowledge Nugget: RLHF and LLMs- The Magic Sauce behind ChatGPT
If you've been intrigued by generative AI and natural language processing, you might have come across two buzzwords: Reinforcement Learning from Human Feedback (RLHF) and Large Language Models (LLMs). But what do these terms mean? And how do they come together to create advanced models like ChatGPT?
In this insightful article,
provides a deep dive into the concepts and their synergy. It answers when RLHF will improve your model’s performance (refer image below). It also addresses the issue of hallucinations in LLMs and proposes solutions to mitigate them.
Why does this matter?
An accurate understanding of how RLHF empowers LLMs’ performance helps build more capable, context-aware, and reliable AI systems. And the knowledge about the issue of hallucinations helps design better reward functions and improve the training process.What Else Is Happening❗
📸Introducing DreamGenerator, a camera that transforms your photos with generative AI! (Link)
🗞️OpenAI commits $5M to the American Journalism Project to support local news (Link)
🌆Google is testing AI-generated Meet video backgrounds (Link)
🤝McKinsey partners with startup Cohere to help clients adopt generative AI (Link)
💰SAP invests directly in three AI startups: Cohere, Anthropic, and Aleph Alpha (Link)
💼Lenovo unveils data management solutions for enterprise AI (Link)
🚀Nvidia accelerates AI investments, nears deal with cloud provider Lambda Labs (Link)
🛠️ Trending Tools
EditAir: AI/ML tool to convert long videos into shareable clips for social media with speech transcription and facial detection.
Visual QR: Generate beautiful QR codes using AI with templates or create your own unique codes.
Sune: AI-first workspace to bring docs, sheets, projects, and tasks to life in a new way.
AI Icebreaker Generator: Generate original icebreaker questions for meetings, workshops, and team-building activities based on purpose, names, and time.
Proddy: AI assistant that turns brief ideas into extensive product docs, saving time.
Intervy GPT: Personal AI learning assistant for teams to learn about any topic and each other.
Printwith AI: Generate AI images and bring them to life on merch with ChatGPT.
Sidekic AI: AI assistant to save, tag, organize and share web resources in starterpacs.
That's all for now!
If you are new to ‘The AI Edge’ newsletter. Subscribe to receive the ‘Ultimate AI tools and ChatGPT Prompt guide’ specifically designed for Engineering Leaders and AI enthusiasts.
Thanks for reading, and see you tomorrow. 😊