selcuka

5 ms·

61.
by selcuka 2mo ago
Another cause of coding harness inflation is that every model provider release their own coding agent, optimised for their models.
62.
by selcuka 2mo ago
> why are ai companies shredding books instead of using Anna's archive? I'm pretty sure they've already consumed everything on Anna's Archive, The Pirate Bay etc., and they are now buying books to fill the gaps and to
63.
by selcuka 2mo ago
The author tested in on an M5 laptop too: > It feels pretty slow on both the M5 Mac and the DGX Spark.
64.
by selcuka 2mo ago
> It's also unclear wherever the subscription price is the real cost, or the API. Define cost: Is it only the inference cost to the provider, or do you also consider training costs as well? If it's the latter, how would you est
65.
by selcuka 2mo ago
> vi editor [...] you had to exit it I didn't know that was possible.
66.
by selcuka 2mo ago
NodeJS is just an example that must have seemed reasonable to the author at the time this blog post was written (11 years ago). Obviously it may no longer apply. Also the definitions of "unknown" and "boring" are always
67.
by selcuka 2mo ago
I'm confused. Your example contradicts your point. How is it a good filter for fraud if it can be aced by only studying for 3 weeks?
68.
by selcuka 2mo ago
But the source codes of those web apps will also be watermarked. /s
69.
by selcuka 2mo ago
Dupe: https://news.ycombinator.com/item?id=49250109
70.
by selcuka 2mo ago
> Nginx’s real strength was being able to proxy and cache HTTP using that same mechanism Fun fact: Igor Sysoev originally wrote mod_accel [1], an acceleration and reverse-proxy caching module for Apache before he made nginx. I remember u
71.
by selcuka 2mo ago
That URL works for me. It must be an issue on your end.
72.
by selcuka 2mo ago
It gave the correct answers to both questions for me: > Bruce Lee was born in San Francisco, California, USA on November 27, 1940. > Bruce Lee's father was a Chinese opera singer That being said, this is not a good test. It is a
73.
by selcuka 2mo ago
It's a quantised 8B model (Llama 3.1 8B to be exact). [1] https://taalas.com/the-path-to-ubiquitous-ai/
74.
by selcuka 2mo ago
Their PoC chips are big, but then it's ridiculously fast (have you seen chatjimmy.ai?). Also they must be holding a bunch of patents.
75.
by selcuka 2mo ago
No wonder both drug dealers and software developers call their customers "users".
76.
by selcuka 2mo ago
> I'm not sure where to find the game itself these days Here you go: https://torinak.com/qaop/play/chronos
77.
by selcuka 2mo ago
I'm pretty sure one can rent a GPU for a few minutes with the electricity cost of leaving an M5 overnight.
78.
by selcuka 2mo ago
> The little apps i made here and there before AI are being outdone by a bunch of drunk people on a whim! Yes, but the premise of the article is you should be able to outdo a bunch of drunk people with your 30 years experience, if you us
79.
by selcuka 2mo ago
I could never understand what it said when I played that game back in time, but today I learned that it was "Meteor Alert! Meteor Alert!" Quicksilva later improved their speech engine. QS Chess, for example, clearly said "Thi
80.
by selcuka 2mo ago
Deterministic != Correct
81.
by selcuka 2mo ago
> AI is not deterministic, it will be making tons of mistakes. From the paper: > The COBOL source is passed through an internal deterministic Migrator to produce a generated Java target. Also, humans are not deterministic either. Gi
82.
by selcuka 2mo ago
I haven't actually tried it, my comment was based on the "Known bugs and issues" section in the readme (e.g. no MCP support yet). Apologies if that section is out of date. If the functionality is on par with the mainstream on
83.
by selcuka 2mo ago
Fair enough. As someone who was alive in the 80s, I remember first generation home computers (ZX Spectrum, Commodore 64 etc) were (correctly) called 8-bit, and the second gen (Amiga, Atari ST) were colloquially called 16-bit. The CPU was 32
84.
by selcuka 2mo ago
Fasttracker II was a 32-bit DOS executable, but trackers in general (and the MOD/XM formats) were born [1] during 16-bit Commodore Amiga era. [1] https://en.wikipedia.org/wiki/Ultimate_Soundtracker
85.
by selcuka 2mo ago
A harness needs to be good. Being large is annoying, sure, but it's generally not an issue. Unless this one has feature parity with the "large" ones (hint: it doesn't), I don't see any reason to use it only because
86.
by selcuka 2mo ago
I guess usury is analogous to lock-in in your parent comment, not the Christian attitudes.
87.
by selcuka 2mo ago
Inject what exactly? Why should I trust incompleteideas.net more than my ISP or government?
88.
by selcuka 2mo ago
Two proxy models that extend the same base model.
89.
by selcuka 3mo ago
> some times you might want more than one schema for a db table Django has the concept of "proxy models" where you can define a trimmed down model for the same database table.
90.
by selcuka 3mo ago
You can also have a /models/ package with foo.py, bar.py under it. It is easier to maintain. As an additional bonus, you can add an __init__.py that imports all models if you want to be able to do a `from .models import FooModel`.

More ›