Hacker News: fumeux_fume

New comment by fumeux_fume in "Travel locally, where you are"

fumeux_fume — Thu, 11 Jun 2026 23:46:12 +0000

I couldn't help reading this article in the voice of Ted Flanders. It's such an insipid point and the writing is so bad. Not the writers fault, but when you find out he lives in Switzerland it's kinda hilarious.

New comment by fumeux_fume in "Disagreement Among Frontier LLMs on Real-World Fact-Checks"

fumeux_fume — Thu, 28 May 2026 15:26:58 +0000

This is a good pattern because it would allow all the models to "think" a bit before giving an answer even if they don't have reasoning or thinking turn on. Just make sure you have the reasoning output before the final answer. A mistake I see all the time is having the answer outputted first then the explanation after which leaves more room for models to rationalize bad answers.

Good pattern: {"explanation": , "answer": }

Bad pattern: {"answer": , "explanation": }

New comment by fumeux_fume in "Disagreement among frontier LLMs on real-world fact-checks"

fumeux_fume — Thu, 28 May 2026 15:14:25 +0000

Do you understand how problematic this is?

New comment by fumeux_fume in "Disagreement among frontier LLMs on real-world fact-checks"

fumeux_fume — Thu, 28 May 2026 13:49:17 +0000

I think we can all agree that this experiment being flawed in multiple ways is TRUE. But I think it's a great exercise in identifying common mistakes people make when using LLMs. This would be a great interview question for a prompt engineering job.

New comment by fumeux_fume in "Magnifica Humanitas"

fumeux_fume — Tue, 26 May 2026 03:14:33 +0000

> But I think the Catholic church should start thinking about an answer to that.

Questions like these about the soul are not new. Religious philosophers have been thinking about them for hundreds of years.

New comment by fumeux_fume in "Hacker News front page as a site"

fumeux_fume — Mon, 25 May 2026 23:54:10 +0000

Nice! Happy to see the site appearing as itself on the front page doesn't cause some crazy recursive crash :)

New comment by fumeux_fume in "Jensen–Shannon Divergence"

fumeux_fume — Mon, 25 May 2026 22:32:48 +0000

There are many applications. I mainly see it used for detecting drift in datasets for ML models. It has a nice benefit over the KL divergence in the case where the two distributions you're measuring have no overlap (KL won't compute, but JS will just return 0). Also, when taking its square root you get a distance rather than a divergence which allows you to compare it to JSD measurements of other distributions.

New comment by fumeux_fume in "US Government releases first batch of UAP documents and videos"

fumeux_fume — Fri, 08 May 2026 14:30:22 +0000

Crackpots,psyops and honeypots, oh my!

New comment by fumeux_fume in "Another Day Has Come"

fumeux_fume — Thu, 23 Apr 2026 04:06:40 +0000

When you read stuff like "existential grief" you gotta roll your eyes.

New comment by fumeux_fume in "The Harvard Library Passport"

fumeux_fume — Wed, 08 Apr 2026 14:51:02 +0000

Ok Deckard

New comment by fumeux_fume in "The Harvard Library Passport"

fumeux_fume — Wed, 08 Apr 2026 13:58:11 +0000

> For those looking for a more discreet way to spend quality time, my sources tell me the single-occupancy office rooms on the top floor are sometimes left unlocked.

This was an interesting little rundown on all the different libraries at Harvard made all the more enjoyable by the author's humor and wit.

New comment by fumeux_fume in "Founder of GitLab battles cancer by founding companies"

fumeux_fume — Sat, 28 Mar 2026 18:28:22 +0000

The linked post about his treatment is basically a vanity article; low in useful information, but high in vague assertions and platitudes. There's also a link to a post griping about the red tape someone experienced while trying to self-treat their dog's cancer that's weird. I clearly live in a different world than these people.

New comment by fumeux_fume in "Bayesian statistics for confused data scientists"

fumeux_fume — Sun, 22 Mar 2026 05:11:55 +0000

As a data scientist, I find applied Bayesian methods to be incredibly straightforward for most of the common problems we see like A/B testing and online measuring of parameters. I dislike that people usually first introduce Bayesian methods theoretically, which can be a lot for beginners to wrap their head around. Why not just start from the blissful elegance of updating your parameter's prior distribution with your observed data to magically get your parameter's estimate?

New comment by fumeux_fume in "Bayesian statistics for confused data scientists"

fumeux_fume — Sun, 22 Mar 2026 04:46:41 +0000

Given your bias, why bother making this point on a thread about using Bayesian methods where they are applicable? Just seems like unconstructive negativity.

New comment by fumeux_fume in "A Japanese glossary of chopsticks faux pas (2022)"

fumeux_fume — Sat, 21 Mar 2026 02:44:09 +0000

My heart is lightened to learn inserting the chopsticks into your mouth to make walrus fangs is not taboo.

New comment by fumeux_fume in "New iPad Air, powered by M4"

fumeux_fume — Mon, 02 Mar 2026 15:53:31 +0000

FWIW, the iPad Air I bought a couple years ago has a small protrusion for the single camera lens, but does not wobble when laying flat and is not really noticeable. This latest iPad Air has a similar design.

New comment by fumeux_fume in "Just two days of oatmeal cut bad cholesterol by 10%"

fumeux_fume — Sun, 01 Mar 2026 03:38:41 +0000

I make it on the stove top. It also takes 30 minutes to cook and 10 to cool. My ratio of water to dry oatmeal is 3 to 1 by weight in grams and I mix in honey after it's done.

New comment by fumeux_fume in "I am directing the Department of War to designate Anthropic a supply-chain risk"

fumeux_fume — Fri, 27 Feb 2026 23:56:03 +0000

Working with the government is typically a huge pain in the ass unless you have a lot of friends on the inside. It's not hard to do the math when you you dealing with a government whose acting incredibly oppositional.

New comment by fumeux_fume in "An Introduction to the Codex Seraphinianus, the Strangest Book Ever Published"

fumeux_fume — Fri, 27 Feb 2026 14:45:27 +0000

I sold my first edition almost 10 years ago to fund (partially) my unemployment during a career transition to data science. A couple years, ago my brother bought me a nice reissue for Christmas without ever knowing I once owned a copy. Odd how some things will make their way to you in the world.

New comment by fumeux_fume in "Gemini 3.1 Pro"

fumeux_fume — Sun, 22 Feb 2026 04:41:54 +0000

Got a free Gmail account? Add $20 or so and you'll be swimming in Gemini outputs. Yet both companies also have a cumbersome onboarding process if all you want to do is get an API token. So yeah, quite similar!