kolinko

4 ms·

1.
by kolinko 2mo ago
What else?
2.
by kolinko 2mo ago
IIRC Claude reads PDFs natively, or as images. One page is 3k tokens IIRC, so you can get 100 pages into context and still have plenty of space to reason or work around that. Only for longer and text-specific documents it uses pdftotext, bu
3.
by kolinko 2mo ago
What do you consider "moderately complex software"?
4.
by kolinko 2mo ago
What you said doesn't disagree with what the parent said. LLM can be trained on a code and at the same time reproduce the core ideas. That's what LLMs do after all - they convert the training data into their own internal models an
5.
by kolinko 2mo ago
> you should not trust it's "opinions" and "knowledge" How do you recognize someone as having true opinions and knowledge from someone having just appearance of them?
6.
by kolinko 2mo ago
Love it, reminds me of early days of crypto. How do you decide upon a price?
7.
by kolinko 2mo ago
If you have more test data I can try this with my harness - you can email me at kolinko@gmail.com btw :) Thanks for the benchmark, I was seriously looking forward to it! Would you consider such even results good for a one-shot? I wonder how
8.
by kolinko 2mo ago
Bombed country? South Korea? Would be Vietnam but US lost there. Nonbombed countries - a ton, Ukraine for one, but the whole post-communist block.
9.
by kolinko 2mo ago
From where in the rest of the world are you? Did even you talk to anyone living in countries neighboring Russia? Sure not everyone benefitted from US but a ton of people/countries did and still do. US never felt a need to build walls t
10.
by kolinko 2mo ago
Here’s the oneshot, not sure if it’s slop or not though :) https://kolinko.eu/pdf-reading-order/ But I wonder about your opinion.
11.
by kolinko 2mo ago
Thanks :) Another cool book about back and forth between monopolies and decentralisation in our space: https://www.amazon.com/Master-Switch-Rise-Information-Empire... It also has an audiobook.
12.
by kolinko 2mo ago
As a general rule, if another country has an open war with your neighbor and openly says that you may be next, don’t put in private data into the sites controlled by that country.
13.
by kolinko 2mo ago
It has steep costs, but with monopolists they have other ways of extracting value from the inventions. Awesome book about the history if Bell Labs - virtually all semiconductor tech we use today was created there (transistors, ics, solar, l
14.
by kolinko 2mo ago
I think it’s the other way around - why should you care about a monopolist’s interests?
15.
by kolinko 2mo ago
Becuse they are a monopolist and we can require certain things from monopolists. Similarly, Bell Labs was kind of required to release transistor for anyone to license - it was a part of social contract that they were allowed to maintain the
16.
by kolinko 2mo ago
I’d expect everything that you type into Yandex that can be of use to the Russia will be used by Russia - they will nit care about hurting you. Just like Russia doesn’t care about online criminals being located within their borders - as lon
17.
by kolinko 2mo ago
A nice story, but I don’t think there really was a time where closed source models’ development stalled really.
18.
by kolinko 2mo ago
Yeah that’s why heuristics should work on the lowest possible layer, not on pdftotext. If you use pdftotext you’re stripping positional data and other stuff. Do you use a public set of documents? I bet I could almost oneshot this with my ha
19.
by kolinko 2mo ago
I think when it comes to precision movement we never went back and we probably won’t go back. We’ll surely have different ways to accomplish it, but society would need to collapse big time for us to not be able to recreate mechanics to oper
20.
by kolinko 2mo ago
Yes of course
21.
by kolinko 2mo ago
Executive branch in Poland and I guess the rest of Europe as well. Bureau for Consumer Affairs. They make decisions, give penalties to corps and invalidate harmful ToS. Then if a corporation doesn’t like it (and they never do), they can app
22.
by kolinko 2mo ago
There are governments with working institutions believe or not. Also, I just read up and in our legal system a nonindividual agreement is nonbinding if “it is drastically against the consumer’s interests or against ‘good manners’”, with fre
23.
by kolinko 2mo ago
Very small peak at the beginning but yes, it’s mostly a long tail. Not always and not in all fields but in most cases afaik it is how it is.
24.
by kolinko 2mo ago
Oh, I just dropped all the comments. Memory and session histories are for storing these things, comments just add confusion. Subsections - how do they run into the same problem? What falsification? The views I mean are dynamic ad hoc. Every
25.
by kolinko 2mo ago
I would argue that creating an llm book that sells also requires human labour, just a different kind of one.
26.
by kolinko 2mo ago
I gave a blanket ban on pdftotext to my agents. The output can get so mangled that a smart human wouldn’t untangle it. Did you try understanding the output from pdftotext yourself? My approach is just ocr-ing with Terra or Gemini flash + ch
27.
by kolinko 2mo ago
What harness? And why not just Fable+Opus? Personally I use only one family for code editing, other families for code navigation. Models tend to reason about the code using their own intuition and putting other families onto the same codeba
28.
by kolinko 2mo ago
Who says anyone needs to hold a mental model? Imho a better way of thinking about it is in terms of views - you can have an unimaginably complex codebase that nobody reads in full, but either agents or humans receive views/perspectives
29.
by kolinko 2mo ago
first you’d need to define how to define/measure the complexity of code when it comes to this case.
30.
by kolinko 2mo ago
It’s pretty well established by now that harness matters as much as a model. Even if a model is smart enough to work around harness’ limitations, it may take it way more time, context and money.

More ›