Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
wrxd
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
7 ms
·
91.
▲
by
wrxd
5mo ago
I'm not sure I understand how this work https://huggingface.co/google/gemma-4-E4B-it-assistant has 78.8M parameters while the standard variant https://huggingface.co/google/gemma-4-E4B-it has
92.
▲
by
wrxd
5mo ago
Since last summer a lot of flags appeared all over the UK. https://en.wikipedia.org/wiki/Operation_Raise_the_Colours https://manchestermill.co.uk/the-men-who-raised-the-flags/
93.
▲
Servers with Personality
(caolan.uk)
2 points
by
wrxd
5mo ago
|
0 comments
94.
▲
by
wrxd
6mo ago
I use Vaultwarden because I need something cross-platform (macOS and Linux) and with support for passkeys. The idea of using different storage methods for different categories of passwords is very interesting. Maybe I don’t need rbw (the ru
95.
▲
Approaching Zero Bugs?
(daniel.haxx.se)
13 points
by
wrxd
6mo ago
|
3 comments
96.
▲
by
wrxd
6mo ago
Unlimited SMS isn't the solution. In fact most of the countries where Whatsapp is dominant already have unlimited SMS (probably because pretty much nobody is using it anymore). What I'd like to see is interoperability so that I am
97.
▲
by
wrxd
6mo ago
Is the installation only imperative via the docker cli tool or do you support something more declarative like Docker compose?
98.
▲
by
wrxd
6mo ago
Agency is part of the reason why I am exploring local models. I don’t particularly like that we get only the weights but even that is much better than depending on an API
99.
▲
by
wrxd
6mo ago
To be fair that’s a Google convention. Have a look at z-pages
100.
▲
by
wrxd
6mo ago
Tangential. This is a cool website, so cool that I tried to subscribe to it in my RSS reader… and it didn’t work. If any of the authors read this message, please consider adding a RSS feed and you’ve got a subscriber!
101.
▲
by
wrxd
6mo ago
What does it mean that they’re running a test? If you’re one of the unlucky 2% you need to pay more?
102.
▲
by
wrxd
6mo ago
This is a risky move. I might have paid $20/month for my personal projects but the Max subscription is a bit steep. Now I'm going to learn more about local models. I don't need to be as good as a frontier model. Good enough a
103.
▲
by
wrxd
6mo ago
I saw this coming when they stopped allowing downloads of purchased books. When that happened my Kindle got KOreader and after it broke I replaced it with a Kobo. I’m much happier with my new device
104.
▲
by
wrxd
6mo ago
You can transfer over USB until you have to factory reset the device. At that point it won’t re-activate and you’ve got a nice brick
105.
▲
by
wrxd
6mo ago
Said that, if you are not in the business of self-hosting, you can easily load books via USB or and it even has Dropbox or Google Drive integrations that work out of the box if you prefer network synchronisation
106.
▲
by
wrxd
6mo ago
It works well and I haven’t had any issue. It doesn’t get full marks because it’s not an open device itself but more a device that’s easy to hack. For example, most books in the kobo store have DRM (to be fair I don’t think it’s entirely th
107.
▲
by
wrxd
6mo ago
What does it even mean?
108.
▲
by
wrxd
6mo ago
I’m so happy I downloaded all my Kindle books when I still had a chance and then moved to the Kobo ecosystem, which albeit not perfect is much much better
109.
▲
Sandboxes and Worktrees: My Secure Agentic AI Setup in 2026
(mikemcquaid.com)
2 points
by
wrxd
6mo ago
|
0 comments
110.
▲
by
wrxd
6mo ago
Same here. I really hope in a near future local model will be good enough and hardware fast enough to run them to become viable for most use cases
111.
▲
by
wrxd
6mo ago
I read good things about https://omlx.ai but I don't really know enough of the ecosystem to know if there are better options. If someone has opinions please let us know!
112.
▲
by
wrxd
6mo ago
I like this direction. The approach of letting LLMs figure out stuff on their own is flexible but inefficient and getting more structure can be very beneficial. Well done for actually parsing the shell commands the LLM wants to execute. Ano
113.
▲
OMLX – LLM inference, optimized for your Mac
(omlx.ai)
2 points
by
wrxd
6mo ago
|
1 comments
114.
▲
Maki – The Efficient Coder (Agent)
(maki.sh)
5 points
by
wrxd
6mo ago
|
0 comments
115.
▲
by
wrxd
6mo ago
This looks cool. Does it support working with local models?
116.
▲
by
wrxd
6mo ago
It’s still capped at £1/kWh, which is a lot but also a price I’ve seen maybe for a few slots on a couple days in the last three years. The price was known ~1 day in advance so I had the choice to fill up the battery with cheaper (but s
117.
▲
by
wrxd
6mo ago
The framing of the article is very odd. It says that Ubuntu increase the requirements not because of the OS itself but to have a better user experience when people have many browser tabs opened. Then it compares to Windows which has lower n
118.
▲
by
wrxd
6mo ago
I hope this means the years social media companies spent optimising for engagement are starting to backfire. > A central issue driving the data is the changing nature of social media itself. The rise of apps such as TikTok and the popula
119.
▲
The Last Quiet Thing
(terrygodier.com)
2 points
by
wrxd
6mo ago
|
0 comments
120.
▲
by
wrxd
6mo ago
Partition Magic was indeed magic. Thanks for the great work you did on it!
More ›