- NATURAL 20
- Posts
- Easier Coding with Meta's Code Llama
Easier Coding with Meta's Code Llama
Meta's latest AI innovation, designed to simplify coding and debug human-written work, all while boasting impressive benchmark performance.
Wes Roth
August 25, 2023
Today:
Meta launches own AI code-writing tool: Code Llama
Meta's got a new toy called Code Llama that'll help folks polish up their coding. Emilia David from The Verge tells us that this tool can spit out new code and fix mistakes in what we've already written. It's free, whether you're just messing around or using it for business.
You can tell Code Llama what you want, and it'll whip up the code for you. They've even got special versions for Python and ones that can understand plain English. But heads up – not all versions of Code Llama play nice with each other.
Meta's bragging that Code Llama outperforms other similar tools, scoring pretty well in some coding test. They're releasing it in three sizes, with the smallest one working super fast on standard gear.
OpenAI names Scale AI ‘preferred partner’ to fine-tune GPT-3.5
OpenAI is spicing things up for its GPT-3.5 Turbo model, letting users customize it with their own data. To make it even cooler, they're teaming up with Scale AI. The two are like peanut butter and jelly – OpenAI has this awesome model, and Scale AI is the pro at fine-tuning it.
What's Scale AI's secret sauce? Their Data Engine. It not only helps fine-tune the model but speeds up its learning. They're not newbies either; they've been helping other models step up their game for a while.
Brad Lightcap over at OpenAI says partnering with Scale AI is a no-brainer. They've got the know-how to help businesses squeeze the most out of OpenAI's models.
They've been showing off their skills with a company called Brex. Brex wanted their expense memos to be top-notch, and they tapped into the powers of GPT-3.5 with a little help from Scale AI. The verdict? Fine-tuning GPT-3.5 is like giving it superpowers, making it almost as cool as GPT-4 but way cheaper.
LinkedIn’s Feed Is Getting Smarter Thanks To AI
LinkedIn's jazzing up their game with some AI magic. They're making the feed smarter so it knows what you wanna see. Think of it like having a pal who knows exactly what stories to tell ya after a long day. For folks using LinkedIn to network or find jobs, this means the posts you see are gonna be more up your alley.
This smarter feed will help get your message to the right folks at the right time. If you're trying to sell, hire, or just get your brand out there, LinkedIn's got your back. With LinkedIn's AI getting sharper, y'all need to up your game. Focus on real engagement, not just big numbers. Maybe even try out some chatty ads. And remember, LinkedIn's gonna keep tweaking this, so stay on your toes and keep testing your stuff.
Amazon is bringing a whole lot of AI to Thursday Night Football this season
Amazon's spicing up "Thursday Night Football" this season with some high-tech AI magic. If you catch a game on Amazon Prime Video, expect some fancy on-screen insights. Think, during nail-biting moments, a graphic might pop up showing if a team should make a risky move, or which player is open and ready for the ball. They're also adding neat features like guessing if a defender's about to make a surprise move, based on 35,000 plays they fed into their computer.
Last season, Amazon just wanted to stream football without a hitch. This year? They're going big on AI to make viewers feel like insiders, breaking down the nitty-gritty of the game.
Amazon's even rolling out a "Prime Vision" stream for the hardcore fans who want all that juicy data. But if you're just there for a regular game, Amazon's got you covered with their basic stream, and they're even bringing back some fan-favorites for additional commentary.
Modular secures $100M to build tools to optimize and create AI models
Modular, a startup aimed at making AI easier to use and more efficient, just raked in $100 million in funding. They're planning to expand their products and their team, but they're not just throwing that cash at fancy computers.
Founded by ex-Google folks, Modular is on a mission to un-complicate the messy world of AI tech. They've got an engine that can make AI models run up to 7.5 times faster, and they’re developing a programming language called Mojo that aims to combine the best parts of existing languages.
While the big tech giants are wrestling with the skyrocketing costs and complexities of AI, Modular says they've got a more affordable, simpler solution. They've already got a growing community of developers and a long waitlist of companies wanting in, so they could be onto something big.
South Korea’s Naver launches generative AI services to compete with ChatGPT
South Korea's tech giant Naver is jumping into the generative AI arena, trying to take on established players like OpenAI's ChatGPT. Known as the "Google of South Korea," Naver has launched its own large language model called HyperClova X. It's designed to enhance the search, marketing, and shopping experiences for users and advertisers alike.
What's interesting is that Naver is aiming to go beyond mere conversation. Their Clova X chat app, also powered by HyperClova X, offers features like document summarization and language translation to help increase work productivity. They're claiming time savings of 10% to 50% on tasks like these, which if true, could be a real game-changer in enterprise settings.
AI Startup Hugging Face Valued at $4.5 Billion After Raising Funding From Google, Nvidia
Hugging Face, an AI company, just got a massive cash boost of $235 million, pushing its worth to a cool $4.5 billion. Big tech names like Google, Amazon, and Nvidia threw their money into the pot.
So, what's the big deal about Hugging Face? Since 2016, they've been the go-to spot for folks to drop AI models for others to grab and use. Think of it like a community library for AI. They also cook up their own AI stuff. With this fresh money, they're planning to hire more brainiacs to join their crew, which already numbers 170.
The boss man, Clement Delangue, says they're in it for the long game, aiming for another decade of growth. Oh, and they're not just hoarding AI models. They've got over half a million of those and 250,000 datasets. They let you grab 'em for free but make their bucks by charging for the extras like more computer power and better customer help. Right now, they've got 10,000 folks happily paying for those perks.
Loftie “Magic Storymaker” - A Groundbreaking AI-Powered Personalized Bedtime Story Experience
A lot of folks in the U.S. are missing out on some serious Zzzs, and it's costing us a ton – like over $411 billion a year. Not enough sleep can lead to big issues like car crashes and health problems. But hey, sleep isn't just about avoiding the bad stuff; it's our body's recharge button. Think of it like sipping on a power smoothie for your brain.
Loftie's stepping into the spotlight with a solution. This company, started by Matthew Hassett, originally wanted to help folks cut down on phone time. They pivoted to reinvent the alarm clock, which honestly hasn’t had a major glow-up since the 80s. The new Loftie Clock helps you chill out before bed with wellness content, white noise, and even custom bedtime stories.
Ryan Rhodes, the tech guru behind Loftie, helped it scale big time. Over 80k customers are now using this nifty gadget, and those bedtime stories? Total hit.
What's cool is that Loftie uses AI to whip up unique stories for users. You toss in some details, and boom – a bedtime story just for you, read in a voice that doesn't sound all robotic.
US Judge Rules AI-Generated Art Not Protected by Copyright Law
A judge in Washington, D.C. said AI-made art ain't gonna get copyright protection because it's missing the human touch. This is the first time a US court has laid down the law about copyright for AI art. Judge Beryl Howell backed up the US Copyright Office when they turned down computer whiz Stephen Thaler's request to copyright a piece made by his fancy AI, "Creativity Machine."
Thaler's company, Imagination Engines, uses artificial brain-like tech. He took the Copyright Office to court last year after they said "no dice" to his copyright ask for a picture that looks like train tracks under a green stone arch. Thaler was all, "My computer did it by itself!" But the office was like, "Copyright's for stuff made by people, man."
Two-thirds of professionals think AI will increase their skills' value
Despite all the chatter about AI taking jobs, most pros are actually feeling pretty good about it. Thomson Reuters asked 1,200 folks from North America, South America, and the UK about AI and their work. Get this: 64% think AI will make their skills more valuable in the next five years. That's a glass-half-full kind of view, right? On the flip side, 66% don't see AI putting them out of a job.
Now, what scares folks the most about AI isn’t getting fired (19%) but actually AI messing up and giving wrong info (25%). Like, imagine using a chat tool for some facts and it feeds you a bunch of baloney. That’s the worry. One lawyer even said clients might get the wrong end of the stick using AI for research.
The cool part? Most of those surveyed (68%) see more work being done by folks without the fancy diplomas. Plus, they think AI will open up brand new gigs. We're already seeing some new jobs, like folks getting paid to come up with lines for chatbots. So, AI? More like A-Interesting.
What'd you think of today's edition? |
What are MOST interested in learning about AI?What stories or resources will be most interesting for you to hear about? |
Reply