Catchmaker

Looking for a job at a top AI company? Catchmaker.com is your new starting point. We just soft-launched in public beta. Feedback welcome!

@VictorTaelin Nicely done.

I *WAS* WRONG - $10K CLAIMED! ## The Claim Two days ago, I confidently claimed that "GPTs will NEVER solve the A::B problem". I believed that: 1. GPTs can't truly learn new problems, outside of their training set, 2. GPTs can't perform long-term reasoning, no matter how simple it is. I argued both of these are necessary to invent new science; after all, some math problems take years to solve. If you can't beat a 15yo in any given intellectual task, you're not going to prove the Riemann Hypothesis. To isolate these issues and raise my point, I designed the A::B problem, and posted it here - full definition in the quoted tweet. ## Reception, Clarification and Challenge Shortly after posting it, some users provided a solution to a specific 7-token example I listed. I quickly pointed that this wasn't what I meant; that this example was merely illustrative, and that answering one instance isn't the same as solving a problem (and can be easily cheated by prompt manipulation). So, to make my statement clear, and to put my money where my mouth is, I offered a $10k prize to whoever could design a prompt that solved the A::B problem for *random* 12-token instances, with 90%+ success rate. That's still an easy task, that takes an average of 6 swaps to solve; literally simpler than 3rd grade arithmetic. Yet, I firmly believed no GPT would be able to learn and solve it on-prompt, even for these small instances. ## Solutions and Winner Hours later, many solutions were submitted. Initially, all failed, barely reaching 10% success rates. I was getting fairly confident, until, later that day, @ptrschmdtnlsn and @SardonicSydney submitted a solution that humbled me. Under their prompt, Claude-3 Opus was able to generalize from a few examples to arbitrary random instances, AND stick to the rules, carrying long computations with almost zero errors. On my run, it achieved a 56% success rate. Through the day, users @dontoverfit (Opus), @hubertyuan_ (GPT-4), @JeremyKritz (Opus) and @parth007_96 (Opus), @ptrschmdtnlsn (Opus) reached similar success rates, and @reissbaker made a pretty successful GPT-3.5 fine-tune. But it was only late that night that @futuristfrog posted a tweet claiming to have achieved near 100% success rate, by prompting alone. And he was right. On my first run, it scored 47/50, granting him the prize, and completing the challenge. ## How it works!? The secret to his prompt is... going to remain a secret! That's because he kindly agreed to give 25% of the prize to the most efficient solution. This prompt costs $1+ per inference, so, if you think you can improve on that, you have until next Wednesday to submit your solution in the link below, and compete for the remaining $2.5k! Thanks, Bob. ## How do I stand? Corrected! My initial claim was absolutely WRONG - for which I apologize. I doubted the GPT architecture would be able to solve certain problems which it, with no margin for doubt, solved. Does that prove GPTs will cure Cancer? No. But it does prove me wrong! Note there is still a small problem with this: it isn't clear whether Opus is based on the original GPT architecture or not. All GPT-4 versions failed. If Opus turns out to be a new architecture... well, this whole thing would have, ironically, just proven my whole point 😅 But, for the sake of the competition, and in all fairness, Opus WAS listed as an option, so, the prize is warranted. ## Who I am and what I'm trying to sell? Wrong! I won't turn this into an ad. But, yes, if you're new here, I AM building some stuff, and, yes, just like today, I constantly validate my claims to make sure I can deliver on my promises. But that's all I'm gonna say, so, if you're curious, you'll have to find out for yourself (: #### That's all. Thanks for all who participated, and, again - sorry for being a wrong guy on the internet today! See you. Gist: gist.github.com/VictorTaelin/8…

@KTmBoyle Agreed. For insiders, it's higher risk and potentially high reward, so can be worthwhile (and more fun/interesting) as long as you have financial cushion to survive being pushed out.

One of the greatest ways to learn is to be so far outside the status hierarchy of a place you can ask the hard and dumb questions no one else can. The outsider advantage is less about having nothing to lose and everything to gain (though that’s part of it.) It’s more that an outsider can operate adjacent to hierarchy and learn faster because they’re not encumbered by it.

Many people try various ways to look smart, without realizing that the opposite is a critical part of high achievement: "I could try out lots of really "dumb" ideas very fast"

Outstanding thread.

@icreatelife Genius. marta-puerto.com Note to #JobSeekers: yes, it's frustrating. But there are many ways you can stand out. Here's one.

Marta was laid off and she made this video to market.. herself. She got a lot of interviews since she posted it. Lots of inspiration in this video for anyone interested in marketing.

@silentworks @supabase That was OAuth with Flask; here's an independent example in the Pyramid web framework: twitter.com/MakeDeeply/sta… ... open to feedback

@_Catchmaker @supabase I've finally gotten around to the OAuth with Python post twitter.com/silentworks/st…

Using @supabase with Python Flask has never been easier. Define your Flask session storage and then create your Supabase client. Learn more from our Python docs supabase.com/docs/reference…

@benjedwards @bilawalsidhu BeOS from @gassee may have sufficed from a technical sense (and was in the running) but WWW was born on NeXT @timberners_lee, and totally agree that the 'learn and grow' was vital to Jobs' future success. As does JLG twitter.com/gassee/status/…

@bilawalsidhu No because he needed that decade to learn and grow—and develop an OS that would later help save Apple. Apple 2.0 would not have happened without that growth period

This whole Sam OpenAI fiasco makes me wonder if social media was a thing during the time of Steve Jobs ousting - could he have avoided a decade long exile?Traditional media has too many layers of abstraction between the actual will of the people. On X - it’s playing out in realtime.

@shreyas Agreed. Great advice for #JobSeekers. The company doesn't care why you need a job, or how badly you need it -- they may have >100 candidates with equally compelling stories. Instead, understand what they're looking for and show how you're the best fit.

When you’re looking for a job, there is a big difference between coming across as self-confident (improves your chances) and coming across as someone who deserves the job (ruins your chances esp. for coveted jobs). Remember: You need the job. The company doesn’t owe you the job.

@samokhvalov @PostgreSQL Peak so far! But plenty of room to grow.

Is @PostgreSQL at its peak of popularity if there is still room to grow?

@jeffreyhuber The essay is from 6 years ago: Nov 11, 2017. Not sure whether still ahead of the curve, or describes a path that won't play out as forecast. Still worth reading and pondering, including for possible affects on career.

i tell almost every person i meet to read karpathy.medium.com/software-2-0-a…

Lots of people post visual market maps but many of us want the underlying data. Here's a real example x.com/omooretweets/s… that has several challenging aspects. Goal: html table with 50 rows: number, cropped logo, company name, domain, and both favicon and logo extracted from home page -- useful in general and important as a cross-check. Willing to give it a try?

If you're looking to parse images, text, tables and more from documents, we got you. Any document, any layout, any file type. We can do it.

@patricksmalone Investment in @Terray_Tx from NVIDIA! And 19 #JobOpenings across bio, chem, data, software & more. "Terray is a biotechnology company with the technology, data, and mindset to radically change the way we discover and develop small molecule therapeutics."

the founding thesis of Terray Therapeutics was that the biggest bottleneck in AI-driven drug discovery is a paucity of high-quality data at scale. five years in, Terray has built an end-to-end platform that integrates automated chemical experimentation and generative AI, all purpose-built with the end (generating safe, effective, drug-like small molecules) in mind. excited to announce a collaboration and investment in Terray from NVentures, @nvidia's venture arm. Terray and NVIDIA will collaborate on completing SOTA chemistry foundation models, some of which will be made available on NVIDIA's BioNeMo platform for AI in drug discovery. by open-sourcing some of Terray's models, the drug discovery community will see what we see: the "unreasonable effectiveness" of high-quality experimental data in solving complex problems in drug discovery.

"Maybe so, maybe not. We’ll see."

Upvoted! Echoing my comment here: "Good to have a small, agile competitor to DocuSign. (I worked with a team that added enhanced e-signature to an existing SaaS app. I don't know anything about this particular team, but it's certainly feasible for a small company to build a compelling solution in this space.)"

Instead of a thousand words — we're on Product Hunt right now, and we could really use your support! At Sendforsign, we simplify digital contracts🤌🏻 If you have a moment, please visit our launch page. Your support would be greatly appreciated🖤 producthunt.com/posts/sendfors…

@itamar_mar @github @Replit @amasad @CodiumAI One way to differentiate: support BBEdit and other desktop editors/IDEs. A smaller but passionate market.

.@github just tried to kill my startup (codium.ai) and many others ‣ example #1: GitHub Mobile ⚔️ @Replit (@amasad) ‣ example #2: Copilot for PR & Chat ⚔️ @CodiumAI Git & IDE plugins ‣ example #3: Copilot Chat for JB ⚔️ cursor.so BUT .!.

@ryanwinchester Great thread, including for #JobSeekers Plua a special note to #HiringManagers - it's fine to hire people who have had the same title for years, even decades. And people who want to move from management back into an individual contributor role.

If I plateau at Senior Software Engineer is that okay? I’ve turned down Lead. I don’t want to deal with people more. I’m bad at it. I want to write code. Mentor a bit. What’s wrong with that?

@WindwardAI @Milipol_Paris A conference. And 7 #JobOpenings in London, Tel Aviv, and Wash. DC. windward.ai/company/career… "the leading Maritime AI™ company, offering a decision support platform to accelerate global trade. Stakeholders can make real-time, predictive intelligence-driven decisions ..."

Explore the future of #maritimesecurity with us at @Milipol_Paris! Join us as we showcase the most innovative solutions to protect our maritime borders & territorial waters. Visit our booth to see how our Maritime AI™ tech is setting new standards in global security. #Milipol

@tabnine Funding! And 7 #JobOpenings - tabnine.com/careers "a pioneer and market leader in ... AI-Driven software development. As one of the few Enterprise-Grade solutions in this space, we have more than 1 million developers who regularly use our product"

📢 We're thrilled to announce that Tabnine has successfully raised $25 million in new investment to continue our mission to accelerate and simplify the work of developers and engineering teams while maintaining the utmost respect for privacy and security. 🎉 We extend our heartfelt thanks to all of you who have been part of our journey. Together, we are actively shaping the future of software development. ✨💜 #GenAI #TechNews #SoftwareDevelopers tabnine.com/blog/tabnine-s…

@TechCrunch Funding! And 17 #JobOpenings in engineering, operations, and more. boards.greenhouse.io/maymobility "May develops and deploys autonomous vehicles (AVs) powered by our innovative Multi-Policy Decision Making (MPDM) technology that literally reimagines the way AVs think."

This startup is riding the autonomous vehicle wave while Cruise and Waymo feel the heat. techcrunch.com/2023/11/07/may…