Amazon to Invest $4B in Anthropic
Plus: Meta's sassy chatbot for young users, LongLoRA for efficient fine-tuning of long-context LLMs.
Hello Engineering Leaders and AI Enthusiasts!
Welcome to the 111th edition of The AI Edge newsletter. This edition brings you Amazon’s $4B investment in Anthropic, expanding access to safer AI.
And a huge shoutout to our incredible readers. We appreciate you! 😊
In today’s edition:
💰 Amazon to Invest $4B in Anthropic
🤖 Meta to develop a ‘sassy chatbot’ for younger users
🚀 LongLoRA: Efficient fine-tuning of long-context LLMs
📚 Knowledge Nugget: RAG vs. Finetuning LLMs - What to use, when, and why. by
Let’s go!
Amazon to Invest $4B in Anthropic
Amazon will invest up to $4 billion in Anthropic. The agreement is part of a broader collaboration to develop the industry's most reliable and high-performing foundation models.
Anthropic’s frontier safety research and products, together with Amazon Web Services’ (AWS) expertise in running secure, reliable infrastructure, will make Anthropic’s safe and steerable AI widely accessible to AWS customers. AWS will become Anthropic’s primary cloud provider for mission-critical workloads, and this will also expand Anthropic’s support of Amazon Bedrock.
Why does this matter?
It will enable enterprises to build with Anthropic models on Amazon Bedrock, responsibly scaling the adoption of Claude and delivering safe AI cloud technologies to organizations worldwide.
Meta to develop a ‘sassy chatbot’ for younger users
Meta has plans to develop dozens of chatbot ‘personas’ geared toward engaging young users with more colorful behavior. It also includes ones for celebrities to interact with their fans and some more geared towards productivity, such as to help with coding and other tasks.
Why does this matter?
Reportedly, Meta is also working on developing a more powerful LLM to rival OpenAI. Perhaps this could serve as a stepping stone towards more advanced AI capabilities and also boost engagement on Meta’s social media platforms.
LongLoRA: Efficient fine-tuning of long-context LLMs
New research has introduced LongLoRA, an ultra-efficient fine-tuning method designed to extend the context sizes of pre-trained LLMs without a huge computation cost.
Typically, training LLMs with longer context sizes consumes a lot of time and requires strong GPU resources. For example, extending the context length from 2048 to 8192 increases computational costs 16 times, particularly in self-attention layers. LongLoRA makes it way cheaper by:
1. Using sparse local attention instead of dense global attention (optional at inference time).
2. Using LoRA (Low-Rank Adaptation) for context extension
This approach seems both easy to use and super practical. LongLoRA performed strongly on various tasks using LLaMA-2 models ranging from 7B/13B to 70B. Notably, it extended LLaMA-2 7B from 4k context to 100k and LLaMA-2 70B to 32k on a single 8x A100 machine, all while keeping the original model architectures intact.
Why does this matter?
LongLoRA is an important step toward making model expansion more computationally efficient. For those interested in creating open-source LLMs with longer context lengths, LongLoRA may be the lowest barrier to entry.
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: RAG vs. Finetuning LLMs - What to use, when, and why
RAG (Retrieval Augmented Generation) and finetuning are two popular methods for using LLMs with “custom” data. However, it can be confusing to know which method to use, when, and why.
In this insightful article,
Clarifies that RAG and finetuning are fundamentally different tools for different problems. (includes a table comparing the two)
Lists out the right use cases of RAG and finetuning.
Lists out other factors to consider when considering RAG and finetuning.
Presents a set of heuristics for choosing what method to use and when.
Why does this matter?
The article helps AI developers navigate between the two methods and avoid analysis paralysis and premature optimization. Moreover, it assists enterprises in making informed investment decisions by clarifying when and how to apply these methods effectively.
(Source)
What Else Is Happening❗
📱Microsoft’s mobile keyboard app SwiftKey gains new AI-powered features
It will now include AI camera lenses, AI stickers, an AI-powered editor, and the ability to create AI images from the app. (Link)
📸Google Pixel 8’s latest leak shows off big AI camera updates
AI photo editing with Magic Editor will enable you to remake any picture you take. DSLR-style manual camera controls will let you tweak the shutter speed and ISO of an image and a focus slider. (Link)
🤖A drinks company in Poland appoints AI robot as 'experimental’ CEO
Dictador, best known for its rums, has appointed the robot to oversee the company’s growth into one-off collectables, communication, or even strategy planning. It is named Mika. (Link)
🎙️ElevenLabs launches free book classics narrated by high-quality AI voices
It presents 6 classic stories told by compelling AI voices in multiple languages, including "Winnie the Pooh" and "The Picture of Dorian Gray." The entire recording process took only one day. (Link)
💡Salesforce to acquire Airkit.ai, a low-code platform to build AI customer service agents
The GPT-4-based platform allows e-commerce companies to build specialized customer service chatbots that can deal with queries around order status, refunds, product information, and more. (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. 😊