netdur

7 ms·

1.
by netdur 26d ago
no worry, name is intentional, it's C + packages, language itself is tiny, strictly bounded by the C ABI, and doesn't include most of the features you'd expect in a language, even standard library is just a external package,
2.
by netdur 27d ago
I am guilty in a sense, I started a new programming language called C+ and one of the pillars I set was that it should not be an alternative to C (or any other language), in fact it uses Clang as its compiler, most of the points in the arti
3.
by netdur 1mo ago
like finding a rock of gold underground and tossing it away because it’s dirty?
4.
by netdur 1mo ago
Remember, this is HN, behave like hackers
5.
by netdur 1mo ago
unlike many of the commenters, I found this pretty nice and educational, there are full stack developers who cannot write anything outside React and do not even understand why React was created in the first place
6.
by netdur 2mo ago
why would any software want to have Kubernetes moment? can't count how devop I know that is confused by it
7.
by netdur 2mo ago
Do something instead of complain
8.
by netdur 2mo ago
The only problem I had with Qwen, fine tuning on Colab, it takes 31 t/s while Gemma 4 is around 9 t/s, otherwise, one of best local LLM
9.
by netdur 2mo ago
Perhaps never, Claude will just be replaced by something else, and eventually we’ll have the Amish of AI
10.
by netdur 2mo ago
basically investors interested in long term and see how spacex the only one in game
11.
by netdur 2mo ago
very narrow vision, openai and anthropic realized what they have probably won’t lead to agi so they moved the goalposts to replacing jobs, programming just happened to be the easiest field because engineers are technical, willing to pay, an
12.
by netdur 2mo ago
Zig and Rust solve different problems, Bun got burned by its indiscipline, and Zig’s response was sadly to flood HN with “how to use Zig” articles
13.
by netdur 2mo ago
Here’s the source of the problem https://en.wikipedia.org/wiki/Grand_Ethiopian_Renaissance_Da...
14.
by netdur 3mo ago
Google is the only LLM frontier that can supply huge enterprise grade AI, yet still struggle, the other one is spacex but their LLM is Grok
15.
by netdur 3mo ago
Why do you want to tax openai on anthropic's fud mistake?
16.
by netdur 3mo ago
It will happen, it's just now is not best time for it
17.
by netdur 3mo ago
Anthropic would never provide weight to anyone for local hosting
18.
by netdur 3mo ago
Because TPU has 2GB and weight + context needs more
19.
by netdur 3mo ago
Yes, it is issue of scale, google had to restrict usage because hardware are not available, regardless of what kind of hardware that is
20.
by netdur 3mo ago
That is one of best answers
21.
by netdur 3mo ago
not sure if I understand you, but 4Q and QAT 4Q are different
22.
by netdur 3mo ago
had a good run with Gemma 4 E2B Unsloth 4Q: https://youtube.com/shorts/XLsAnz5aAAI The E4B model doesn’t fit on my phone TPU, so it swaps to RAM, the QAT version means more accuracy, good!
23.
by netdur 4mo ago
That's great, spotting Opus text got easier
24.
by netdur 4mo ago
I find helpful ads on Google Search sometimes, and it can be the easiest way to get results, but most of the time, ads (and SEO) ruin search accuracy to the point that it's becoming totally useless
25.
by netdur 4mo ago
I am working https://vibu.app which is free digital voucher and for fun, I am building yet another programming language!
26.
by netdur 4mo ago
Mistakenly, i thought it was about Rotten Tomatoes, and i started thinking about how a movie like Michael ranked badly, the critics missed the whole point of watching a movie, to be entertained, sadly, here on HN, sometimes we miss the poin
27.
by netdur 4mo ago
Then do not use Android, some Chinese phone or iPhone
28.
by netdur 4mo ago
I am getting 21 t/s on Fold 7, 21 x 1.8 = 37.8 t/s compared to M1 Max's 54 t/s, that is impressive
29.
by netdur 5mo ago
why would the bbc publish an article about a man who believed a chatbot's story about someone coming to kill him?
30.
by netdur 5mo ago
hmm... at Q4_K_M, stock-style quantization is retaining ~99–99.8% of BF16 accuracy, AutoRound pushes that to ~99.4–100.n% (??) the gap is roughly 0.1–0.7 percentage points https://github.com/intel/auto-round/blob&#

More ›