@NFLonFOX It's a failed experiment that shouldn't have made it to air. In the score bug, the team name is far too big. The font weight is heavy to the point of obesity requiring 1px kerning to even fit. You can't fix bad typography by just super-sizing it. Visually fatiguing!
Today we're announcing the winners of ARC Prize 2024. We're also publishing an extensive technical report on what we learned from the competition (link in the next tweet).
The state-of-the-art went from 33% to 55.5%, the largest single-year increase we've seen since 2020. The benchmark remains unbeaten, but we're happy to see that research progress on the key bottleneck to AGIs (in particular on-the-fly adaptation to novel tasks) has been reignited in 2024 -- in part thanks to ARC Prize.
In particular the competition has popularized Test Time Training (TTT), originally pioneered for ARC-AGI by Jack Cole last year. I believe TTT represents the largest jump in LLM generalization capabilities since the initial findings regarding in-context-learning circa 2019-2020. ARC Prize has also led to a considerable surge of research interest towards program synthesis.
Competition winners:
🥇 the ARChitects (Daniel Franzen, Jan Disselhoff)
🥈 @guille_bar
🥉 alijs (Agnis Liukis)
Paper Award winners:
🥇 "Combining Induction and Transduction For Abstract Reasoning" by @xu3kev et al.
🥈 "The Surprising Effectiveness of Test-Time Training for Abstract Reasoning" by @akyurekekin et al.
🥉 "Searching Latent Program Spaces" by @ClementBonnet16 & @MattVMacfarlane
ARC-AGI-Pub Leaderboard (solutions using commercial APIs):
🥇 @jeremyberman
🥈 @ellisk_kellis & @akyurekekin
🥉 @RyanPGreenblatt
$44,000 H100 GPU Black Friday Deal - $10 off! But I'm pondering the "Returnable until Jan 31, 2024" offer. Potential business model? Asking for a friend... @beffjezos@OpenAI
Instead of "I’m sorry, Dave. I’m afraid I can’t do that," we're going to get a long, vague, and conspicuously even-handed discourse about the factors involved in the design and use of pod bay doors.
✨ I made my first video game with ChatGPT:
1) ChatGPT generates a text-based adventure game with DALL-E 3 generating images for it
2) Every time you play the game is different because it generates the story and images live
3) The images from DALL-E are sent to @runwayML which turns images into video
4) The text is sent to @elevenlabs which turns the text adventure into a pirate narrator voice
5) It's merged into a video
6) Interactive buttons are overlayed
The game is called:
🐒🏝️🇳🇱The Secret of Monkey Island: Amsterdam (unofficial)
And you can play it here:
https://t.co/fhaX667hTJ
(video + TTS + buttons doesn't work auto yet, for now manual but text + img works, I'm building an interface for it now)
@zarehgorjian@Visible@Verizon@visiblesupport Thanks Zareh. Reaching out on Twitter helped break through the logjam at @VisibleCare and connect me directly with a senior tech who resolved the unusual system bug I'd been caught in within a few hours. All good now!!!
Help! A week ago I switched to @Visible phone service. For 7 days now, no one can call or text me & @Visible is holding my 30+ yr phone # hostage. Trying to port out my # to @Verizon but @Visible Error. @VisibleSupport admit it's their prob but can't fix. >9 hrs wasted on chat!
Help! A week ago I switched to @Visible phone service. For 7 days now, no one can call or text me & @Visible is holding my 30+ yr phone # hostage. Trying to port out my # to @Verizon but @Visible Error. @VisibleSupport admit it's their prob but can't fix. >9 hrs wasted on chat!
I am such a fan of @ProfEmilyOster and her team. Full time in-person school = best for kids, best for control of COVID. Win-win. Been saying this since last summer but never had data this robust to support it.
Also, no reason to keep forcing kids to mask. Landmark study.👏
24% increase in youth suicides in CA since 2019, when the trend had previously been improving. Absolutely heartbreaking but not one bit surprising. Watched my own kids turn into shells of their former selves with school & sports closures. Suicide more risky for kids than COVID.
Big news from the CDC: If you’re fully vaccinated, you do not need to wear a mask – indoors or outdoors, in most settings.
We’ve gotten this far. Whether you choose to get vaccinated or wear a mask, please protect yourself until we get to the finish line.
Within 10-20 years, nearly every branch of science will be, for all intents and purposes, a branch of computer science.
Computational physics, comp chemistry, comp biology, comp medicine... Even comp archeology. Realistic simulations, big data analysis, and ML everywhere
@Boris I love this! What a fun idea @Boris!
Building a virtual email Lenny sounds like a perfect job for an AI text generation engine like GPT-3. There are enough formal business email replies online to build quite a training corpus with minimal effort. Ideal student project.
@KostyaMagician @MagicLive2019 Thanks! Thinking about a rigorously deep innovation framework to "imagine new impossibilities" for magician/inventors was both challenging and a refreshing departure from tech/business creativity. If you see me around pls stop and say hi. I'd like to hear your perspective.
At our workshop at @facebook in Chicago, today @markran shares his journey in creating #Kickboxorg, the world’s most popular enterprise #innovation framework
@toddrundgren Driving by my local theater today I saw on the billboard that you're playing here on Thursday! DM if you're up to grab coffee and catch up.