HNHacker News
TopNewBestAskShowJobs

vwkd

28 karma · joined December 4, 2023

submissionscomments
vwkd··on Asciinema 3.0 will be rewritten in Rust
Asciinema is great! The only thing it’s lacking is full screen support for mobile. Currently, on mobile the full screen button doesn’t work. I can zoom in the page but this never really aligns well.
vwkd··on Deno in 2023
Here's the specific talk https://youtube.com/watch?t=822&v=dipwQfcV0AU (CascadiaJS channel) or https://youtube.com/watch?t=822&v=5DX49vzLfqw (Deno channel)
vwkd··on Anki – Powerful, intelligent flash cards
Interesting question. Good memory seems certainly beneficial. To me an even more interesting question is "memory of what?". What things are worth remembering? Is it worth to remember all 300 local plant species?

Maybe biology had good reasons to not evolve us with perfect memory. I found that over time I only remember the big picture and forget the details that make it up. And mostly that's fine! Usually, I only need the big picture. Like a math theorem, where once you understood it it's fine to forget the details of the proof since going forward you'll only need the big picture of the theorem.

Of course, sometimes life is not a simple straight line and might circle back to ask again for previous details. But then, how many details, how often does it happen? Maybe it's fine to forget and look it up again in those few cases?

Personally, I found little need for active memorization in real life. If something is worth remembering, it usually comes up multiple times, and after looking it up repeatedly, it tends to get remembered automatically.

Sure, sometimes real life poses constraints in time or exposure that hinder natural memorization (e.g. educational exams, remembering names, etc.) in which case active memorization may be useful. But these occasions seem to happen rather rarely over the whole course of a life.

For these reasons, I don't use Anki personally. Instead, I try to accept having to look things up again when it happens, if necessary repeatedly. No, it's not easy to not be disappointed at myself that I forgot again. But we can't avoid the pain - it's always there in some form. Either in the forgetting, or in the work to not forget.

https://news.ycombinator.com/item?id=39172246

vwkd··on Meta AI releases Code Llama 70B
> when investigating models and evaluating interview candidates

Wow, just realized, in the future employers will mostly interview LLMs instead of people.

vwkd··on Meta AI releases Code Llama 70B
What’s the advantage of Together? The price is about the price of GPT 3.5 Turbo ($1/mil tokens is $0.001/thousand tokens), which has the advantage of wide ecosystem and support.
vwkd··on "If nothing changes, all remaining Nitter instances will go down eventually"
Care to elaborate for those not familiar with kiwifarms?
vwkd··on "If nothing changes, all remaining Nitter instances will go down eventually"
With an AWS IP and a bot usage pattern they’ll surely ban your account pretty quickly or put you in front of a CAPTCHA. I wish it was as easy as a small script. Without anti-bot techniques, sites would be overflown by scraping bots. Try to scrape a Cloudflare protected site, for example. They’re really good in figuring out if you’re human or a bot. IIRC they even fingerprint your TLS handshake or cypher suite, which ultimately made me give up with headless Chrome and Puppeteer even after proxying through my residential IP, spoofing user-agent and screen size and rate limiting. Unfortunately, there’s no way to distinguish good bots for personal usage from bad bots.
← PreviousPage 2 of 2