Hacker Newsnew | past | comments | ask | show | jobs | submit | learningmore's commentslogin

“However, with the same model, if you just use the completions API and then parse the output, it will return the correct quantity…”


The article includes a youtube video, where an NPC shares facts about the archeological site (i assume backed by knowledge from an LLM).

Pretty cool, I can see this as useful to a lot of different museums or organizations.

It would be interesting for the NPC to ask the player how they wanted information to be delivered- “I’m a hurried. Tourist, keep your responses short”.


Nice breakdown.


“We uncover a systematic failure: LLMs cannot correct errors in their own outputs while successfully correcting identical errors from external sources - a limitation we term the Self-Correction Blind Spot.”


FYI, ATS stands for Applicant Tracking System.


correct, thank you


“The Capability Hardware Enhanced RISC Instructions (CHERI) project is a rethinking of computer architecture in order to improve system security.”


Published in 2014.


>Published in 2014.

Yes, that is in the article.

Anything you wish to add?

Developments since then the article lacks?


It's common practice here to point out the date of older articles so the year can be added to the post title.


>It's common practice here to point out the date of older articles so the year can be added to the post title.

I see, I hope you can understand where I was coming from, seeing the year in the title paired with what looks like a snarky comment devoid of any analysis of said article.


it's not a blog article about outdated technical topics, so who cares


It's just a convention. We (moderators) append the year of an article (in parens) when an article is from a previous year. Of course we miss many cases, and commenters often helpfully point those out. In this case colinprince added the year to the title (thanks!) but otherwise we would have.

It's not that anybody did anything wrong—historical material is welcome here! and it's nice for readers to know roughly what time an article dates from. That's all.


Sorry I was snippy, I thought they were casting aspersions on the science inside the article due to the date, which is a pet peeve given anything biosciences related moves much more slowly than CS given the (rightful and good) restrictions around human subjects research.


Application error: a client-side exception has occurred (see the browser console for more information).


Ah, I ran into a gunicorn issue with timeouts and worker timeouts. Thanks for that.


Very interesting article. Well-explained.


Yes! I’ve even considered finding some extension to just accept every cookie option for every site, so just I’m not hammered with consent dialogs on StackOverflow.

I use multiple browsers for testing, across multiple computers.

Perhaps this fact, along with their consent expiration policy causes the consent dialogs to always show.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: