nsagent

28 ms·

1. 301.▲
   by nsagent 3y ago

   I haven't heard any real details of what Q* is supposed to be; I assumed it's related to reinforcement learning. I guess it's possible the heuristic in A* was replaced with a neural network value function. I'm not sure h
2. 302.▲
   by nsagent 3y ago

   Please constructively add to the conversation rather than trotting out poorly reasoned quips. Clearly, natural language is not formal logic [1]. And even if it was, the statement doesn't necessarily translate into formal logic so strai
3. 303.▲
   by nsagent 3y ago

   I agree with the OP, that it is a form of gaslighting. I haven't read the fact checking websites in a while, but seeing them now it's clear that they exhibit political bias just like most news organizations these days while having
4. 304.▲
   by nsagent 3y ago

   There are even more gems in the paper, like this one: > Suppose a leader pledges during a campaign to provide humanitarian aid to a stricken nation or the CEO of a company commits publicly to register its algorithms or guarantee its cust
5. 305.▲
   LitMT: A platform for LLM translations of novels into over 20 target languages(twitter.com)

   2 points by nsagent 3y ago|0 comments
6. 306.▲
   by nsagent 3y ago

   Read my comment [1] for this one academic's take on intelligence in current LLMs. I feel it passes the spirit of the test you provide (doing it literally would be a bit nonsensical). [1]: https://news.ycombinator.com/it
7. 307.▲
   by nsagent 3y ago

   I've seen you frequently argue similar points about LLMs, but matching a statistical distribution of text is imitation, which necessitates confabulation on the part of LLMs. Namely, humans have underlying causes that affect their writi
8. 308.▲
   by nsagent 3y ago

   I'm honestly surprised no one has mentioned the Most Favored Nation clause, which helps enforce Steam's ability to charge 30%. https://www.pcgamer.com/lawsuit-claims-valve-is-abusing-its-... https://www
9. 309.▲
   by nsagent 3y ago

   I'm not so sure about this. I'm not saying courts in the US will rule one way or the other, I'm just saying it's certainly not a forgone conclusion that training is fair use. Even if it is, the companies might not have s
10. 310.▲
    by nsagent 3y ago

    I commiserate, since I too started my PhD later in life, but luckily I don't have any kids and my wife can work remotely. I'm moving for a postdoc right after the Spring semester ends. I understand why the bias against not moving
11. 311.▲
    by nsagent 3y ago

    Early adopters can upgrade MacPorts before it officially supports the latest OS by building from source if they have the patience and are willing to debug any broken ports.
12. 312.▲
    by nsagent 3y ago

    Agreed. I switched from Homebrew to MacPorts a few years ago and couldn't be happier. I just delay upgrading macOS until MacPorts officially supports the latest OS (which often takes a bit longer than Homebrew). NOTE: I briefly tried G
13. 313.▲
    by nsagent 3y ago

    Sadly, if you have an iPhone I don't think you can easily read Nota docs currently. The article that introduces Nota [1] has only been tested on Chrome. I tried multiple browsers on iOS to no avail (likely since they all use the same u
14. 314.▲
    by nsagent 3y ago

    Exfiltrating 300TB of customer data without 23andMe noticing seems like negligence. Shouldn't there have been some alarms?
15. 315.▲
    by nsagent 3y ago

    > On the infrastructural disadvantage, that’s true, but we also have the advantage of … we don’t have to be full of shit. We’re not actually a surveillance company. I’m not trying to pretend Facebook is good. I don’t have to toe a party
16. 316.▲
    by nsagent 3y ago

    Yes, but the algorithm is optimizing for clicks, not curation. Curation is a much more costly process requiring humans in the loop, which often leads to cries of censorship (e.g. my otherwise popular article is being unfairly down-weighted)
17. 317.▲
    by nsagent 3y ago

    Ironically, the title of the article is literal clickbait. The first four paragraphs (out of 21) of the article are related to the title. The majority of the article (11 paragraphs) are dedicated to AI safety. I avoid companies like Google
18. 318.▲
    by nsagent 3y ago

    Maybe don't assume that PhD-level NLP researchers are out of touch on cognitive neuroscience topics related to language understanding. The latest research seems to indicate that language production and understanding exist separately fr
19. 319.▲
    Spin the SDF URL Russian Roulette Chamber(sdf.org)

    2 points by nsagent 3y ago|0 comments
20. 320.▲
    by nsagent 3y ago

    As an PhD student in NLP who's graduating soon, my perspective is that language models do not demonstrate "reasoning" in the way most people colloquially use the term. These models have no capacity to plan ahead, which is a r
21. 321.▲
    by nsagent 3y ago

    > Non-academic jobs and experience - should I include it? > I am a huge supporter of two things not related to research and education: having a job while an undergraduate or showing a substantial dedication to an organization. Sadly,
22. 322.▲
    Generative AI at the BBC(bbc.co.uk)

    2 points by nsagent 3y ago|0 comments
23. 323.▲
    by nsagent 3y ago

    he definitely isn't the first one to use the term in that capacity. it's frequently used by cv researchers in relation to super resolution (e.g. just search "face hallucination" on google scholar) see the abstract of thi