nl

17 ms·

421.
by nl 5mo ago
I always wonder what the world is going to be like when these are 2 orders of magnitude faster. Something like https://chatjimmy.ai/ will do 14,000 tokens per second, and it's a completely different experience to what
422.
Anthropic says it's about to have its first profitable quarter(techcrunch.com)
3 points by nl 5mo ago|1 comments
423.
OpenAI Stargate: where the US sites stand(epoch.ai)
4 points by nl 5mo ago|0 comments
424.
by nl 5mo ago
I've been using the FileSystems API a lot lately. It gives a sandboxed directory on the local computer you can read and write to (user decides what rights you get). It's what the web version of VS Code uses to edit local files. It
425.
by nl 5mo ago
> Ed Zitron His math is wrong though. He still claims H100s are worthless but in fact they are worth more now than when they were new. And everything I've read from him is just.. weird? Like he has an anti-AI agenda and he interpr
426.
by nl 5mo ago
That depreciation it too high in practice. 5 year old H100s are now completed depreciated but are being rented out at higher rates than when they were new. > Who is going to rent the GPUs then? I'd LOVE a way to be on the other si
427.
by nl 5mo ago
I don't see how this follows at all? It's true that Anthropic didn't buy as much compute as OpenAI. But OpenAI's compute purchases are one of the largest investments in human history. It's also true that they are no
428.
by nl 5mo ago
Not as simple as that. Meta's well know campaign was actually to make the app stores (and maybe OSes) responsible for age verification, not apps. Google and Apple campaigned to make apps responsible for it.
429.
by nl 5mo ago
> if you want to train your 5T params model like modern small models are being trained (with a thousands time more training tokens than params), that's an enormous training run. Yes it is. Spending $100M on training runs is common,
430.
by nl 5mo ago
I think you could well be surprised by the Sonnet vs Opus bill (assuming you are paying via the API) In my experience Sonnet bills can be higher than Opus because it churns a lot more trying to get things right. Example from my fairly simpl
431.
by nl 5mo ago
Yes it's a big thing that people are slowly becoming more aware of. Nvidia models are even worse than Qwen! https://sql-benchmark.nicklothian.com/#token-efficiency-and-... (mouse over the cells for token counts and cli
432.
by nl 5mo ago
I mean in general I'm pretty doubtful about things he says, but in this he was comparing Grok and it sort of makes sense in the context: https://x.com/elonmusk/status/2042123561666855235
433.
by nl 5mo ago
This is why people use Gemini on OpenRouter. Same model, don't have to deal with Google billing.
434.
by nl 5mo ago
Elon says Opus is 5T (and I would expect he'd know) > It's not that frontier labs can't create a 5T+ parameter model, but they don't have the data to optimize a model of that size. The have plenty if data. They use ve
435.
by nl 5mo ago
I've very confused. The SVGs show the beak, wing, tail, feet and body as though viewed from directly underneath. They look similar to the photo, but meet the instructions better ("from underneath"). What are you expecting exa
436.
by nl 5mo ago
This is because whoever owns Fivethirtyeight now (ABC?) deleted the whole archive of articles on the site.
437.
by nl 5mo ago
On my Agentic SQL benchmark it scores 19/25. That's... mediocre. It means performs worse than 3.1 Flash Lite Preview (22/25), is slower (367s vs 142s) and is more expensive (75c vs 2c). It is outperformed by Gemma4 26B-A4B in
438.
by nl 5mo ago
Interestingly the `o` in GPT-o4 stood for Omni too (which I never realized until yesterday when reading random 3rd party documentation)
439.
by nl 5mo ago
What on earth do you mean? I live near an area with lots of pelicans. If you look up at one flying overhead this is what they look like. Here is a photo for comparison: https://commons.wikimedia.org/wiki/File:American_w
440.
by nl 5mo ago
I've tried Zed and really didn't like it. I like VS Code with the Claude Plugin, and sometimes with the Codex Plugin
441.
by nl 5mo ago
Most of the performance on coding comes from RL, not distillation. Distillation helps with world knowledge and things like that.
442.
by nl 5mo ago
Have you seen a pelican from underneath? There's not much to show!
443.
by nl 5mo ago
Link doesn't work - maybe not public?
444.
by nl 5mo ago
> Try to get the AI to draw the pelican form a very odd angle - like underneath, to the right, one wing extended, one wing not ... 0% chance. Proof by existence? https://gist.github.com/nlothian/50241d34a654fcf0caa28
445.
by nl 5mo ago
> But 'Drawing of a Proper Duck' is almost arbitrary because it may have nothing to do with the 'Specific Duck You Wanted'. That might be the case, but Simon's case "Generate an SVG of a pelican riding a bic
446.
by nl 5mo ago
The difference is very noticeable as your codebase gets bigger and you give higher and higher level tasks. I've certainly had things that Opus fixed using some kind of work around that GPT-5.5 actually solved. And the difference betwee
447.
by nl 5mo ago
OpenAI use(d|s) Stainless too right?
448.
by nl 5mo ago
IF they do individual billing the business doesn't get token reporting
449.
by nl 5mo ago
The model is out here: https://huggingface.co/Efficient-Large-Model/SANA-Video_2B_7...
450.
by nl 5mo ago
Oh misread that sorry!

More ›