nl

36 ms·

691.
by nl 9mo ago
Bluesky driving the most traffic is unexpected. I have to say I've been enjoying it quite a lot lately. It's not as militantly anti everything as it used to be and that's improved it a lot.
692.
by nl 9mo ago
> I’m simply reposting accurate and direct information that is already public and already highlighted by TFA. No you aren't. To quote: > There ought to be social consequences for using machines to mindlessly and recklessly libel
693.
by nl 9mo ago
How is your hit comment any better than the AI's initial post? It lacked the context supplied later by Scott. Your's also lacks context and calls for much higher stake consequences.
694.
DeepMind Aletheia [pdf](github.com)
5 points by nl 9mo ago|0 comments
695.
by nl 9mo ago
> Given that LLMs are just lossless compression machines, I do sometimes wonder how much better they are at compressing plain text compared to zstd or similar. Should be easy to calculate... The current leader on the Hutter Prize ( http:
696.
by nl 9mo ago
I'm talking about your own coin, not chain. ERC20 tokens are part of Ethereum (and yes I realise there are also non ETH based tokens and that the gas cost of Eth makes them attractive etc etc)
697.
by nl 9mo ago
Sure, but your average developer doesn't have a lot of agency in if the US invades another country in order to increase the value of the coin they got for having a PR merged. But with crypto they do. See for example all the BAGS coins
698.
by nl 9mo ago
No, the perverse incentive is that there will be RepoCoin, and the people involved will be incentivized to make the price of that as high as possible.
699.
by nl 9mo ago
> The right script, with the right prompts can be tailored to create a loop, allowing the premium model to continually be invoked unlimited times for no additional cost beyond that of the initial message. Ralph loops for free...
700.
by nl 9mo ago
This thought pattern leads to crypto. In that world there's a process called "staking" where you lock some tokens with a default lock expiry action and a method to unlock based on the signature from both participants. It woul
701.
by nl 9mo ago
Setting up spreadsheets is just programming in an unconventional environment. And it has all the requirement translation requirements of programming
702.
by nl 9mo ago
Huh - yes you are right. I don't know what I was thinking, but it was very wrong.
703.
by nl 9mo ago
Do people care about how many bytes they are sending or receiving? Most people care about getting the right bytes.
704.
by nl 9mo ago
Did you actually read these or just Google for problems? > "Floating-point arithmetic may give inaccurate results in Excel" Well.. yes? That's what floating point numbers do. https://stackoverflow.com/quest
705.
by nl 9mo ago
When you setup a spreadsheet, you choose a column or cell to show the NPV. You use the npv formula on this cell and apply it to inputs from other columns or cells. This process of deciding what data to put in what column and how to apply NP
706.
by nl 9mo ago
> What nl is attempting to do above Um wow talk about bad faith! Actually it's been about 35 years since I read Animal Farm, so I took what the review said as valid.
707.
by nl 9mo ago
> erm theres an NPV function built in excel. Exactly! The LLM just needs to make sure it uses it appropriately. Doing that bit is the non-deterministic part, but the NPV calculation itself is completely deterministic.
708.
by nl 9mo ago
I don't understand what this comment is trying to say. The commentator mannykannot didn't really comment on the politics other than to agree that 1984 is a criticism of Stalism (which... I don't think anyone would argue wit
709.
by nl 9mo ago
The deterministic part (calculations) is done by Excel. The non-deterministic part is turning human instructions ("calculate the NPV over 10 years for X given Y") into Excel. This is already a non-deterministic process (humans a
710.
by nl 9mo ago
Usually I find this kind of variation is due to context management. Accuracy can decreases at large context sizes. OpenAI's compaction handles this better than anyone else, but it's still an issue. If you are seeing this kind of t
711.
Accelerating Scientific Research with Gemini: Case Studies and Common Techniques(arxiv.org)
4 points by nl 9mo ago|0 comments
712.
by nl 9mo ago
Copilot is such a typical MS product. It checks all the correct checkboxes on a feature list in comparison to the competition but it just sucks to use. It's like Sharepoint - the deathpit of all collaborative software
713.
by nl 9mo ago
> you shouldn’t expect it to be fair about any socialist or communist countries, usually classified as brutal dictatorships, The World Fact Book doesn't have this kind of commentary. For example read the entry on North Korea. I&#x27
714.
by nl 9mo ago
> My understanding is that the paper talks about very general principles. This isn't really true. In this case it's specific to NVidia's tensor matrix multiply-add (MMA) instructions, which lets it use silicon that would o
715.
by nl 9mo ago
> truly good (Claude, Gemini) to evolve Claude yes, but Codex is much better than Gemini in every way that matters except speed in my experience. Gemini 3 Flash is an amazing model, but Gemini 3 Pro isn't great. It can do good wor
716.
by nl 9mo ago
> I think it’s more likely that the API prices will decrease over time and the CC allowances will only become more generous. I think this is absolutely true. There will likely be caps to stop the people running Ralph loops/GasTown w
717.
by nl 9mo ago
I'm not an expert on the geography of Spain, and it's rare that I'd defend Grokipedia but in this case I think it is correct. Meseta Central mean central tableland . Segovia is on the edge of the mountain range that surround
718.
by nl 9mo ago
I think there's a difference between people getting a system a d realising it isn't actually what they wanted and "never survive collision with reality". They survive by being modified and I don't think that invalid
719.
by nl 9mo ago
> Consumers have mostly rejected AI. People like to complain about things, but consumers are heavily using AI. ChatGPT.com is now up to the 4th most visited website in the world: https://explodingtopics.com/blog/ch
720.
by nl 9mo ago
Nemo is different to Megatron. Megatron was a research project. NVidia has professional services selling companies on using Nemo for user facing applications.

More ›