Hacker News: crthpl

New comment by crthpl in "A recent experience with ChatGPT 5.5 Pro"

crthpl — Sat, 09 May 2026 15:09:30 +0000

Yes it can! That's the whole point of RL! it generates slightly out of distribution rollouts, and rewards good rollouts to change the distribution of the output

New comment by crthpl in "A recent experience with ChatGPT 5.5 Pro"

crthpl — Sat, 09 May 2026 15:04:29 +0000

where the heck did you get those parameter numbers from?

New comment by crthpl in "Utilyze measures how efficiently your GPU is doing useful work"

crthpl — Sun, 03 May 2026 14:21:09 +0000

are you sure it's 60 fps and not uncapped?

New comment by crthpl in "I am definitely missing the pre-AI writing era"

crthpl — Mon, 30 Mar 2026 17:20:00 +0000

have you tried pangram? it's basically the only good AI detector, and they have nearly 0 false positives

New comment by crthpl in "Claude is a space to think"

crthpl — Wed, 04 Feb 2026 20:22:27 +0000

Anthropic is very big (the biggest AI co?) in B2B, where you don't have ads. Also, if they end up creating a datacenter full of geniuses, ads won't make sense either.

New comment by crthpl in "The Adolescence of Technology"

crthpl — Tue, 27 Jan 2026 19:11:46 +0000

1. you can definitely tell apart an S-curve and an exponential if you look at the derivative(s). AI progress does not seem to be close to the middle of the S-curve. 2. e.g.: a slowdown hasn't shown up in moore's law yet.

New comment by crthpl in "The creator of Claude Code's Claude setup"

crthpl — Wed, 07 Jan 2026 05:44:19 +0000

It is the case that Anthropic employees have no usage limits. Some people do experiments where they spawn up hundreds of Claude instances just to see if any of them succeed.

New comment by crthpl in "Evaluating chain-of-thought monitorability"

crthpl — Mon, 22 Dec 2025 02:24:13 +0000

the chain of thought is what it is thinking

New comment by crthpl in "Claude Sonnet 4.5"

crthpl — Mon, 29 Sep 2025 17:48:46 +0000

The reason they get a perfect score on AIME is because every question on AIME had lots of thought put into it, and it was made sure that everything was a possible. SWE-bench, and many other AI benchmarks, have lots of eval noise, where there is no clear right answer, and getting higher than a certain percentage means you are benchmaxxing.

New comment by crthpl in "AccountingBench: Evaluating LLMs on real long-horizon business tasks"

crthpl — Mon, 21 Jul 2025 18:54:53 +0000

GPT-4o is so far behind the frontier; you shouldn't use it as an indicator of what LLMs are capable of.

New comment by crthpl in "Shipping WebGPU on Windows in Firefox 141"

crthpl — Wed, 16 Jul 2025 10:48:01 +0000

Another reason may be that WebGPU didn't allow for as much optimization and control as Vulkan, and the performance isn't as good as Vulkan. WebGPU also doesn't have all the extensions that Vulkan has.

New comment by crthpl in "Why I don't ride the AI Hype Train"

crthpl — Wed, 09 Jul 2025 12:13:08 +0000

If humans were too complicated to be mathematically modelled, then we wouldn't exist.

New comment by crthpl in "Microsoft is basically discontinuing Windows. What will happen, then?"

crthpl — Sun, 20 Oct 2024 04:20:07 +0000

Why would charging for extended support hurt Microsoft? They "win" if you pay for it, or if you upgrade to windows 11. If you do neither, you'll probably get hacked down the line if you don't airgap, and Microsoft doesn't gain or lose anything (except goodwill, but that doesn't matter with the kind of monopoly they have).

New comment by crthpl in "TCAS averts possible head-on collision in Austin"

crthpl — Sat, 19 Oct 2024 01:26:48 +0000

I think there is a deterministic algorithm (based on GPS position?) for determining who climbs and who descends.

New comment by crthpl in "New exponent functions that make SiLU and SoftMax 2x faster, at full accuracy"

crthpl — Wed, 15 May 2024 21:40:48 +0000

I'm guessing MacOS, Linux, Android, Windows, iOS, and Web?

New comment by crthpl in "The case for never reading the book jacket"

crthpl — Tue, 28 Nov 2023 21:49:58 +0000

The idiom "Don't judge a book by its cover" applies to many situations, but books are not one of them.

New comment by crthpl in "Vulkan Documentation"

crthpl — Wed, 11 Oct 2023 15:49:12 +0000

dawn is the WebGPU backend in chromium, while wgpu is the WebGPU backend for firefox written in Rust. wgpu is seeing a lot of use in non-browser uses; there are some examples on their website.

https://wgpu.rs/

New comment by crthpl in "MP3 vs. AAC vs. FLAC vs. CD (2008)"

crthpl — Sun, 01 Oct 2023 21:08:44 +0000

The 7 million is probably how many different hues we can see. We can see many more different brightness levels.

New comment by crthpl in "New finding boosts Panspermia theory that life on Earth originated in deep space"

crthpl — Sat, 12 Aug 2023 17:09:08 +0000

Not if only one species came to earth.

Framework Laptop 16 Deep Dive – Enclosure

crthpl — Thu, 22 Jun 2023 18:50:35 +0000

Article URL: https://frame.work/ca/en/blog/framework-laptop-16-deep-dive---enclosure

Comments URL: https://news.ycombinator.com/item?id=36436931

Points: 307

# Comments: 232