Hacker News: heyitsguay

New comment by heyitsguay in "Pretraining Language Models via Neural Cellular Automata"

heyitsguay — Thu, 19 Mar 2026 16:02:09 +0000

What were the results of 3d fractal shader pretraining?

New comment by heyitsguay in "A 26-Gram Butterfly-Inspired Robot Achieving Autonomous Tailless Flight"

heyitsguay — Thu, 26 Feb 2026 17:23:33 +0000

Interesting, can you explain why?

New comment by heyitsguay in "[dead]"

heyitsguay — Thu, 12 Feb 2026 15:57:04 +0000

Is it inflammatory if it's true? Seems rather embarrassing (but fitting for 2026 America) to shy away from painful truths because they're uncomfortable.

New comment by heyitsguay in "ChatGPT's porn rollout raises concerns over safety and ethics"

heyitsguay — Mon, 26 Jan 2026 05:21:09 +0000

Moloch!

New comment by heyitsguay in "Why didn't AI “join the workforce” in 2025?"

heyitsguay — Tue, 06 Jan 2026 05:59:56 +0000

Thanks for this answer! I appreciate the clarity, I can see the economic impact for your company. Very cool.

New comment by heyitsguay in "Why didn't AI “join the workforce” in 2025?"

heyitsguay — Tue, 06 Jan 2026 04:14:31 +0000

Agreed. I've never seen a concrete answer with an outcome that can be explained in clear, simple terms.

New comment by heyitsguay in "Scripts I wrote that I use all the time"

heyitsguay — Wed, 22 Oct 2025 20:07:55 +0000

I went through a similar cycle. Going back to simplicity wasn't about laziness for me, it was because i started working across a bunch more systems and didn't want to do my whole custom setup on all of them, especially ephemeral stuff like containers allocated on a cluster for a single job. So rather than using my fancy setup sometimes and fumbling through the defaults at other times, i just got used to operating more efficiently with the defaults.

New comment by heyitsguay in "Abundant Intelligence"

heyitsguay — Wed, 24 Sep 2025 03:15:32 +0000

What problems?

New comment by heyitsguay in "Serverless Horrors"

heyitsguay — Sun, 07 Sep 2025 16:53:22 +0000

Pass a law requiring cloud compute providers to accept a maximum user budget and be unable to charge more than that, and see how quickly the big cloud providers figure it out.

New comment by heyitsguay in "A PM's Guide to AI Agent Architecture"

heyitsguay — Thu, 04 Sep 2025 21:57:01 +0000

What do you do with agents?

New comment by heyitsguay in "Charting Form Ds to roughly see the state of venture capital “fund” raising"

heyitsguay — Thu, 28 Aug 2025 18:46:47 +0000

Are there any? Concretely. Genuinely curious.

New comment by heyitsguay in "Teaching GPT-5 to Use a Computer"

heyitsguay — Mon, 18 Aug 2025 05:35:27 +0000

"Unfortunately, my content guidelines prohibit me from describing my activities with your mother last night"

New comment by heyitsguay in "Tversky Neural Networks"

heyitsguay — Sat, 16 Aug 2025 19:05:13 +0000

It's the same amount of effort benchmarking, just a better choice of backbone that enables better choices of benchmark tasks. If the claim is that a Tversky projection layer beats a linear projection layer today, then one can test whether that's true with foundation embedding models today.

It's also a more natural question to ask, since building projections on top of frozen foundation model embeddings is both common in an absolute sense, and much more common, relatively, than building projections off of tiny frozen networks like a ResNet-50.

New comment by heyitsguay in "Tversky Neural Networks"

heyitsguay — Sat, 16 Aug 2025 18:29:04 +0000

Seems cool, but the image classification model benchmark choice is kinda weak given all the fun tools we have now. I wonder how Tversky probes do on top of DINOv3 for building a classifier for some task.

New comment by heyitsguay in "Diffusion language models are super data learners"

heyitsguay — Sun, 10 Aug 2025 19:29:59 +0000

Do you have references on adaptive methods for image recognition?

New comment by heyitsguay in "Can Large Language Models Play Text Games Well?"

heyitsguay — Fri, 04 Jul 2025 15:08:33 +0000

I've done something along these lines! https://github.com/heyitsguay/trader

The challenge for me was consistency in translating free text from dialogs into classic, deterministic game state changes. But what's satisfying is that the conversations aren't just window dressing, they're part of the game mechanic.

New comment by heyitsguay in "I counted all of the yurts in Mongolia using machine learning"

heyitsguay — Wed, 18 Jun 2025 15:29:46 +0000

Object detectors output detection bounding boxes along with confidence scores. The higher the score, the more confident the model is that the associated bounding box is a correct detection.

When used in applications (like this one), the user typically establishes a confidence threshold and then every detection above that threshold is treated as a positive detection, the rest are discarded. The choice can be arbitrary or (sorta) principled.

New comment by heyitsguay in "Is there a half-life for the success rates of AI agents?"

heyitsguay — Wed, 18 Jun 2025 14:23:45 +0000

I've noticed this a lot, too, in HN LLM discourse.

(Context: Working in applied AI R&D for 10 years, daily user of Claude for boilerplate coding stuff and as an HTML coding assistant)

Lots of "with some tweaks i got it to work" or "we're using an agent at my company", rarely details about what's working or why, or what these production-grade agents are doing.

Strategic Overclaiming of LLM Reasoning Capabilities Through Evaluation Design

heyitsguay — Sat, 07 Jun 2025 19:04:46 +0000

Article URL: https://huggingface.co/papers/2506.04734

Comments URL: https://news.ycombinator.com/item?id=44211809

Points: 1

# Comments: 0

New comment by heyitsguay in "'Spiderweb' drone attack marks a new threat for top militaries"

heyitsguay — Wed, 04 Jun 2025 05:28:46 +0000

Defense needs them everywhere, offense needs one gap in the defense.