msoad

My layman view is that more compute (more reasoning) will not solve harder problems. I'm using those models every day and when problem hits a certain complexity it will fail, no matter how much it "reasons"

I hope the hedge fund shorted NVDA to make some good money along the way too hahaha!

They could make a ton of money shorting NVDA and releasing the paper. The most honest short position ever!

We are at multiple trillion dollar investment territory, all purely based on the idea that "to make AI you need lots of GPU and power"

four GPUs are very convincing indeed! :D

I use yek

https://github.com/bodo-run/yek

Does it work nicely on Linux? I'm very curious about this

All I want is an iPhone Shortcuts script to delete messages like "Hi" and "Hey" from unknown numbers. I get so many of those and having to delete them is a pain.

Shortcuts does not allow deleting messages apparently :(

I saw this earlier. benchmarks are impressive!

Did OpenAI release anything beside this product? Any benchmarks at least to compare?

It feels like OpenAI is betting on the fact that they have a nice UI?!

This is a very uniformed response IMO. S3 seems very niche compared to Node.js compatibility. Not sure why you're attacking me for saying this?

I have not tried Bun yet but the long list of features makes me skeptical that it's all solid and bug-free. I'm wishing to be wrong. I'll give it a spin in a future project.

From a project management perspective I'm a little confused why would you spend time on S3 support while you're still not 100% Node.js compatible. Next.js is a very big ecosystem and if you can get Next.js customers onboard you'll grow much more than supporting S3.

Can someone please do some security research on this to see what it sends when it calls home?

I'm terrified to install a binary like that

I agree with Tailwind's stance on this. You really don't need @apply if you're breaking things down to smaller components. I often see people have things like <ul><li className="long_list_of_classes">text1</li><li className="long_list_of_classes">text2</li>...</ul>. This is where I think we need a linter to warn against things like that. Make those <li>'s a component!

I think everyone who worked at Google in the past has PTSD from having to migrate gRPC schemas. What a mess! Type safety doesn't have to be this costly

no... one more lane will fix the traffic. Truly American approach

Amazing to see how DeepSeek R1 is doing better than OpenAI models with much less resources

if your birth year starts with 2, I can see why you might think that

I am paying for o1 Pro but since Deepseek R1 came out I stopped using it. So there goes $200/mo of their revenue ;)

I know. It is sad. Naming can also be seen as a way of showing respect to a hugely impactful paper if you want to be positive about it.

Here is a conundrum:

With all that AI generated code being pushed, as a leader I wonder which is better? Enforce a ton of e2e so no code that is really well thought through all aspects of the solution can go past CI or does this enable AI to go even crazier and break all sort of best practices to just pass the test?

I'm not trying to be cheeky here. They have literally joked about how TypeScript versions means nothing really. So they can't just announce a new major version and drop enums completely. Maybe with a feature flag this is possible but even then, a fresh tsc --init not supporting enums is not really how TypeScript works

https://ai.meta.com/research/publications/large-concept-mode...

Google Maps is not showing Palestine on the maps app. it shows middle of the ocean when you search for it...

if how us humans reason about things is a clue, language is not the right tool to reason about things.

There is now research in Large Concept Models to tackle this but I'm not literate enough to understand what that actually means...

I'm not saying that never has happened. maybe they trained against openAI models but they are letting anyone to train from their output. I doubt they had access to GPT models to "distill"

No model really can "call home". It's the server running it. Luckily for Deepseek there are other providers that guarantee no data collection since the models are open source

Give me five odd numbers that don't have the letter 'e' in their spelling

Compare the reasoning times!!! 84s vs 342s

R1 (Thought for 84 seconds)

      No odd number in English avoids the letter 'e' in its spelling. The request for five such numbers cannot be fulfilled.

      No standard English spelling of an odd number can avoid “e.” Every odd digit (one, three, five, seven, nine) already includes “e,” so once you build any odd number out of those digits, the spelled‐out form will contain “e.” As a result, there are no such odd numbers in standard English without an “e” in their name.

It already replaces o1 Pro in many cases for me today. It's much faster than o1 Pro and results are good in most cases. Still, sometimes I have to ask the question from o1 Pro if this model fails me. Worth the try every time tho, since it's much faster

Also a lot more fun reading the reasoning chatter. Kinda cute seeing it say "Wait a minute..." a lot

Yeah they should deprecate namespaces and enums in the next major version... oh wait...

TypeScript versioning is literally a joke

by compiling it, which opens a huge can of worms. Deno relies on tsconfig.json configurations for instance

I think they are referring to `class Foo { constructor(private bar: string) }`