3abiton

4 ms·

1.
by 3abiton 7d ago
Why wait, if there is already a similar implementation a year ago, and it's fully open source: dataset, paper, repo etc ... The idea is not new. https://news.ycombinator.com/item?id=49736660
2.
by 3abiton 7d ago
Few years ago, there was an android app (open source from fdroid) that basically provided the model and ability to listen and give you informafion about the birds. It was very fun during hikes with friends.
3.
by 3abiton 7d ago
This is a great evolution in the right direction compared to LLM + pydantic and temperature 0.
4.
by 3abiton 9d ago
Man this looks amazing, more competition is always welcome! The issue for Ai inference is always firmware/software support. Up until recently, that rocm hardware became usable, let alone optimized. Nvidia had a big start, now everyone
5.
by 3abiton 10d ago
I 100% support this "loophole"
6.
by 3abiton 10d ago
Wait a bayesian advocate using a frequentist approach and not having a fight over which branch is supreme. What's happening with the world. I guess all stats united against AI.
7.
by 3abiton 17d ago
I have no doubts about this, even if you ask chatgpt itself to run a security audit, this would have been one of them flags.
8.
by 3abiton 18d ago
Honestly a big part of this is AMD's lackluster strategy to GPU software. To say it's lacking is understatement. At least for non-data center gpus.
9.
by 3abiton 23d ago
I wonder the effect of diary writing on the brain, I see always recommended as either part of the self-improvement space or meditation. But reading your comment, I wonder if it allows to strengthen memories, especially if ones reread it aft
10.
by 3abiton 1mo ago
> trust One interesting aspect ij academic literature about trust, it is highly reverse correlated with diversity. No matter the grouping and size, the more diverse, the less trust within the group.
11.
by 3abiton 1mo ago
It's very much how I've been using claude code in the past 5 month: brainstorm -> generate specs (constraints) -> generate exact plan -> implementation + review -> test / validate. The last stage is the most cruci
12.
by 3abiton 1mo ago
Unfortunately it's not always easy to move away from systemd. I still run void on one of my machines, but aur make things so much easier.
13.
by 3abiton 1mo ago
Even more videos to binge before the holidays end! Thanks for sharing!
14.
by 3abiton 1mo ago
Alright, a new series to binge now.
15.
by 3abiton 1mo ago
I hate to be pedantic but you can finetune a skill to shape the PR message the way you like it. That being said, I did have exactly this issue you mentioned, but the defualt output can always be tuned.
16.
by 3abiton 1mo ago
Nonetheless, it's the year og the linux desktop!
17.
by 3abiton 1mo ago
I think the github readme repo is more for people who heard about it, but you're absolutely right, it is lacking in context and explanation.
18.
by 3abiton 1mo ago
Congrats, I remember seeing this few months ago, and it has improved a lot since then! I haven't tapped into Wolfram / Mathematica since my physics years, but I am curious now to try it again.
19.
by 3abiton 1mo ago
> Same, sad how far Claude has fallen. I still think Claude code patterns are amazing so I just port those. I am curious, is that a plugin, or skill?
20.
by 3abiton 2mo ago
This was so well beautifully written, and poignant for our times. Almost 70 years old paper.
21.
by 3abiton 2mo ago
Gosh this gave a big myspace flashback
22.
by 3abiton 2mo ago
This is so awesome. I wish more realistic problems get solved like this instead of every new gadget getting shoehorned for gaming or porn.
23.
by 3abiton 2mo ago
I second that. I've always enjoyed his work!
24.
by 3abiton 2mo ago
Check out Thingino. It's an open source firmware you can flash on certain off-the-shelf chinese camera (including wyze), to turn them from cloud spyware to actual fully local ip devices.
25.
by 3abiton 2mo ago
I wonder to what extent these habits or quirks are self-learned or "self-imposed" (probably the wrong word). I guess, it gets us in the comfort zone to free up our brain energy to perform more important tasks.
26.
by 3abiton 2mo ago
Could be handicapped version with more limited functionalities. Noentheless, open-weight is open-weight, I just hope the quality is good enough to be sota.
27.
by 3abiton 2mo ago
I miss the days of 4changpt.
28.
by 3abiton 2mo ago
> Yeah, Qwen3.5-122B-A10B-NVFP4 produces better responses than Qwen3.6-27B-NVFP4 (both from unsloth), but I'm mostly using them for programming in various ways, mostly Rust, Clojure, Python and JavaScript, and some translations task
29.
by 3abiton 2mo ago
> in my experience of 1 month daily use, Qwen 3.7 Pro is just unusable. wastes too much time, goes off track, useless stuck loops, cannot debug at all. Deepseek V4 Pro is night-and-day compare to Qwen. actually Qwen models seems the wors
30.
by 3abiton 2mo ago
Also at scale Kimi (or DeepSeek and others) don't have to worry about local deployment. The number of param keeps growing, I highly doubt consumer hardware will catch up. I am just glad Qwen team is still putting up great bangers.

More ›