ljosifov

7 ms·

31.
by ljosifov 8mo ago
Everyone should do the calculation for themselves. I too pay for couple of subs. But I'm noticing having an agent work for me 24/7 changes the calculation somewhat. Often not taken into account: the price of input tokens. To produ
32.
How to Run Local LLMs with Claude Code and OpenAI Codex(unsloth.ai)
2 points by ljosifov 9mo ago|0 comments
33.
by ljosifov 9mo ago
https://ljubomirj.github.io small personal ~/public_html
34.
When competition leads to human values by Beren Millidge [video](youtube.com)
1 points by ljosifov 9mo ago|0 comments
35.
by ljosifov 10mo ago
+1. For every one like the author of the blog post, it's likely to be another one in the opposite direction. But they will be unlikely to write a post about that. I too found weighting 'spend time with human persons v.s. with my
36.
A new way to extract detailed transcripts from Claude Code(simonwillison.net)
3 points by ljosifov 10mo ago|0 comments
37.
by ljosifov 10mo ago
I imagine - and sure hope so - everyone trains on everything else. Distillation - ofc if one has bigger/other models providing true posterior token probabilities in the (0,1) interval (a number between 0 and 1), rather than 1-hot-N tar
38.
by ljosifov 10mo ago
Nah - given the ergonomics + economics, local coding models are not atm that viable. I like all things local even if just for safety of keeping healthy competitive ecosystem. And I can imagine really specialised uses cases where I run an 8B
39.
by ljosifov 10mo ago
Not for SSD specifically, but I assume the compact design doesn't hurt: duckdb saved my sanity recently. Single file, columnar, with builtin compression I presume (given in columnar even simplest compression maybe very effective), and
40.
We present Olmo 3, our next family of open, leading language models(twitter.com)
1 points by ljosifov 11mo ago|0 comments
41.
by ljosifov 11mo ago
So far so good - and I say this as one voting remain. The only gripe I have is that our domestic doomers were even more stupid than the EU ones. Ours were the progenitors of many of EU dumb ideas. So even outside EU, we in the UK not only d
42.
by ljosifov 1y ago
Yeah - IDK why it never makes it to these lists. R too. Matlab being 2D matrix first/default gets it right for me there. IK matrices trivially translate to arrays, still: find 2D to be extra expressive on human level, for zero price p
43.
by ljosifov 1y ago
Yes - so telling hard truths is not for the benefit of the listener, it's for the benefit of the speaker mostly. That's a major point: if I see, but I don't tell, if I have private truths and public lies, it's one small
44.
by ljosifov 1y ago
Iceland - sorry for the confusion. (poor spelling)
45.
by ljosifov 1y ago
Eh - for the world's audiences: EU is not Europe. I geddit how/why why people equate eu == europe - it would simplify things for all, one niggle less to consider. But - it ain't so, for better or worse. There are countries in
46.
by ljosifov 1y ago
I think you are correct there - the majority of the public don't care. They just try to get about doing their daily business and act the best they can under circumstances. So we just click "Accept" to any popup banner make it
47.
by ljosifov 1y ago
I am a little bit worried, for sure. But I think that's small extra risk on my side, for small extra gain for me personally, but large extra gain for the wider group I belong to (ultimately - all of humanity) in the sense of working to
48.
by ljosifov 1y ago
Yes - I meant 'impossible to difficult' to define to all people, at all times. Agree it's easy for me to define how that looks. It doesn't mean that the same is true to you. That's why I went from a very general, to
49.
by ljosifov 1y ago
You know little about me, so it's better to assume less, no? My personal experience with medical data specifically is, that I would have been harmed by obstacles to data sharing that the UK medical system has in place, having not been
50.
by ljosifov 1y ago
It's not a satire, you can check mu comments on this topic easily. I dispute 'most people'. Revealed preferences of most people are that they value their data privacy very cheaply, almost zero. Even one click extra to share t
51.
by ljosifov 1y ago
My experience in the UK medical systems has been the opposite - wrote here https://news.ycombinator.com/item?id=45066321 Google knows what "Home" is for me only in Gmaps, because I went out of my way (put a Label
52.
by ljosifov 1y ago
That's a consideration, for sure. But given the LLM-s have not got the ground truth, everything is controlled hallucination, then - if the LLM tells you an imperfect version of my email or chat, you can never be sure if what the LLM to
53.
by ljosifov 1y ago
I'm worried, it's not like I don't care. For example, I'm worried that Google is such a huge ginormous target, that at some point their Gmail will be broken. At the same time, there are benefits to sharing data. There ar
54.
by ljosifov 1y ago
I'm not 'screwing' anyone. I'm saying - the same way people don't want to have their data used, I DO want my data used. I'm not saying they use YOUR data. I'm saying they use MY data. Likewise, I'm no
55.
by ljosifov 1y ago
I already share lots of my data with Google. I have Gmail where a lot of my online life is reflected. I have Photos, Gmaps, Gdrive. Also Google knows about my YouTube viewing, my Android phone use. So no matter what I say - with my actions,
56.
by ljosifov 1y ago
Yes - my data, not your data. You stay away from my data. I stay away from your data. I don't care about your data. But I do want them to train on my data. And to serve me better. Was disappointed that they didn't do that already.
57.
by ljosifov 1y ago
I know indoctrination well. Reading what you write - I get the impression that you don't know much about indoctrination. But I don't know you, so I allow it that I maybe wrong. You asked "in what universe". I showed you
58.
by ljosifov 1y ago
Whether Google is interested in serving me or not, is not only untestable (i.e. what counts as 'Google', 'interested', and 'serving' there - one could argue to end of time) - but besides the point. I want to be
59.
by ljosifov 1y ago
I see reading comprehension is not something you enjoy to indulge with. These - > utter fantasies like "'no one wants to share their data' is just assumed" or "the defaults are always 'deny everything'&
60.
by ljosifov 1y ago
Maybe in the US. In the UK, I have found obstacles to data sharing codified in the UK law frustrating. I'm reasonably sure some people will have died because of this, that would not have died otherwise. "Otherwise" case being

More ›