- The AI Daily Brief
- Posts
- OpenAI API Announces 16K Context Window and More
OpenAI API Announces 16K Context Window and More
Plus AMD announces superchip for AI and HuggingFace Partnership
Nathaniel Whittemore
June 14, 2023
The AI Breakdown First Five - Wednesday June 14, 2023
Today on the First Five:
5. AMD Announces Superchip and HuggingFace Partnership
4. Andromeda Cluster for Startups
3. Meta Announces New AI Image Generation Model
2. Adobe Generative Color
1. OpenAI API Increases Speed and Expands Context
5. AMD Says Nvidia Doesn’t Get To Have All The Fun
Nvidia has been all over the news lately, what with their supercomputer, new chip and trillion-dollar valuation. Not so fast, says AMD, who announced their own new superchip for AI the M1300x. They also announced a hardware development partnership with HuggingFace. The game is on!
"We want everyone to be able to run their models on AMD hardware" - CEO HuggingFace @ClementDelangue
"Over 5,000 models in the last week." "Want all to be able to optimize on AMD from Instinct to Epyc to Radeon to Ryzen to Embedded" @huggingface
"Best combo of hardware software*— Dylan Patel (@dylan522p)
6:16 PM • Jun 13, 2023
4. Andromeda Cluster to Expand Startup Development
One of the rate-limiting factors for AI startups is compute. Enter former Github CEO Nat Friedman and entrepreneur/investor Daniel Gross. They’ve put together what they call Andromeda Cluster, a 10 exaflop cluster for startups to use for all their delicious experiments.
Daniel and I have setup a cluster for startups: andromedacluster.com
— Nat Friedman (@natfriedman)
4:05 PM • Jun 13, 2023
3. Meta’s New “Human-Like” AI Image Model
Another day, another Meta open source AI release. Today it’s I-JEPA, a model that generates images by leveraging knowledge about the world to fill in gaps. The team at Meta says that this human-like reasoning could solve common problems of AI image generation (like the dreaded extra fingers).
2. Adobe Generative Recolor
Adobe continues its absolute AI reinvention. At their annual Max event in London yesterday, the team made clear how central AI is to the company’s future vision. They also announced another new AI tool, this time in Illustrator. Generative Recolor allows creators to use text prompts to modify vector images — would could be extremely useful for things like branded content varieties.
Make Art Not War: The Battle for Creativity
It's Adobe's annual Max event in London today and @scottbelsky's spotlight on AI paints a clear picture: AI isn't just on Adobe's agenda, it is the agenda.
Adobe has already scored a home run with generative fill in Photoshop, a… twitter.com/i/web/status/1…
— Bilawal Sidhu (@bilawalsidhu)
11:57 PM • Jun 13, 2023
1. OpenAI’s Big API Update
OpenAI has announced a big slate of API updates. Those include Function Calling (which some have likened to plugins for developers); a 16k context version of the newly updated GPT 3.5 turbo; faster speeds, and cost reduction on embeddings and input tokens. Developers. Are. Stoked.
WHOA THE NEW @OpenAI MODELS ARE SO FAST
— Kevin Fischer (@KevinAFischer)
12:29 AM • Jun 14, 2023
Bonus: From 2D to 3D Video
Nascent but lots of possibility!
"DynIBaR by Zhengqi Li et al." is an AI model that can take a regular 2D video clip and create a 3d version. This unlocks a multitude of scene-dependent edits including focus, animating camera position, stabilisation, stereo-3d and potentially relighting.
dynibar.github.io— Ben Ferns (@ben_ferns)
6:45 AM • Jun 13, 2023
The Latest AI Breakdown Episodes
Thanks for reading! Let us know how this edition hit!
How was today's newsletter? |
The AI Breakdown podcast - https://pod.link/1680633614
The AI Breakdown YouTube - https://www.youtube.com/@theaibreakdown
The AI Breakdown Discord - https://bit.ly/aibreakdown
Signing off from the future’s past - NLW