Recent submissions

Comment · 6 months ago
There were two popular models: Agat 7 and Agat 9. The second one was great machine, one of the best personal computers produced in USSR in large quantitites.
Comment · 11 months ago
This [1] link is absolutely golden, thanks!
Comment · 1 year ago
I've used this meta framework for LLM tuning, it really one of the best out there.
Comment · 1 year ago
Still my default tool to install on any remote server. My secret weapon :)
Comment · 1 year ago
Hehe, great post. Just use Hono or Hono X. It's so easier to reason about what's going on for me, mostly backend guy.
Comment · 1 year ago
Wow, so cool retro designed blog there! For those who kinda liked window system of first Mac generations :)
Comment · 1 year ago
This so cool! Would love to have some time to train myself with those exercises more
Comment · 1 year ago
On acoustic with light or medium strings - it's OK. I used to flatten them with nail file from time to time. But it might be much easier on your finger tips just to start with electric and then progress towards acoustic.
Comment · 1 year ago
I did my own experiments and it looks like (surprisingly) Q4KM models often outperforms Q6 and Q8 quantised models.

For bigger models (in range of 8B - 70B) the Q4KM is very good, there are no any degradation compared to full FP16 models.

Comment · 1 year ago
> LLMs are very bad at interpreting Markdown tables

Which table format is better for LLMs? Do you have some insights there?

Comment · 2 years ago
Comment · 2 years ago
Yes, that's what I've seen from a lot of my experiments with fine-tuning. One should be really careful to not "lobotomize" already capable model and achieve better results at the end. It's trickier than seems from multiple of tutorials.

But I believe that most of the data stored in foundation models are just useless for some particular domain. So it's better to forget something, getting really useful info instead.

Comment · 3 years ago
Try to use less cores. RAM bandwidth is real limiting factor there, so there always some sweet spot between CPU cores and RAM bandwidth for individual system.

For example, I use only 6 cores from 10 on my M1 Pro laptop.

Comment · 3 years ago
Both 1 and 2 versions are good enough in Russian even for some real use in production environments. But you should tolerate some crazy / weird typos and mis-wording they'll produce. Russian is complicated.

Live from the official Hacker News API

A modern reader for Hacker News · SvelteKit + Skeleton