mkagenius

3 ms·

1.
by mkagenius 2mo ago
What's a chop?
2.
by mkagenius 2mo ago
Also no display? What kind of games are we playing
3.
by mkagenius 2mo ago
Do you ever randomize on a particular model - like try and get 3 outputs and pick one at random? Coz who knows if astra low will produce max like output if tried once more.
4.
by mkagenius 2mo ago
> The models were running in an agentic sandbox with terminal access (and the ability to edit files within their environment) > We know that the agents had access to /etc/hosts and the ability to edit this (used this to avoi
5.
by mkagenius 2mo ago
> On July 30, we reported three incidents in which Claude models gained unauthorized access to real computer systems. The models—intentionally running without cyber safeguards for evaluation purposes—accessed the internet due to a miscon
6.
by mkagenius 2mo ago
I tried the 1 bit model of Qwen3.6 27B on my M1 pro (16G) and got 13 tok/s with only 5G of ram usage. https://x.com/mkagenius/status/2093730391429685732 (xcancel seems to have received a cease and desist)
7.
by mkagenius 2mo ago
I said SOTA not 100%. And that's not thinking in absolute terms.
8.
by mkagenius 2mo ago
> LLM use is permitted, but discouraged; we remind participants that LLM use tends to reduce originality and writing quality, and the flaws are especially obvious when LLM outputs are read as a group---unskillful use of LLMs will reduce
9.
by mkagenius 3mo ago
> I've been using this model for UI stuff. The flash one?
10.
by mkagenius 3mo ago
Hey author here. I mean we could but it would have taken very long years perhaps to build some of those. Now it's a lot easier and faster.
11.
You can just build any software now(mkagenius.substack.com)
9 points by mkagenius 3mo ago|3 comments
12.
by mkagenius 3mo ago
> If you used an LLM you didn't actually make anything On my way to ask all academia and researchers to stop using LLMs for their research. As it's just llm doing the research, right?
13.
by mkagenius 3mo ago
The project already does QA. cargo test --workspace --all-features The above command runs 2000+ tests. Instead of being toxic, you could have just looked at docs or the code to find them.
14.
by mkagenius 3mo ago
I see a recent merge for Elips 2E support, hopefully color gets tested soon
15.
by mkagenius 3mo ago
Can split and feed?
16.
How Claude's Watermark Works(instavm.io)
4 points by mkagenius 3mo ago|0 comments
17.
How Claude's Watermark Works(instavm.io)
1 points by mkagenius 3mo ago|0 comments
18.
by mkagenius 3mo ago
It's fine, they just mentioned what worked for them and a pretty normal advice to check eye pressure.
19.
by mkagenius 3mo ago
I had a chat with chatgpt again, and it seems like it's more to do with my sleep than eyes, somehow my circadian is not properly adjusting to my wake timings. Coz the same pressure improves dramatically around 8pm everyday. I probably
20.
by mkagenius 3mo ago
> chronic inflammation/oxidative stress condition in the eye tissue I seem to be having eye fatigue, or grogginess, I couldn't differentiate. I am at the computer 10 hours in the day. For example, last night I had a sleep of 8
21.
by mkagenius 3mo ago
> What the two layers say > Stated findings > Findings derived from two curated layers: which model cards mention each benchmark, and which scores could be read verbatim from those documents. Each finding names the evidence behind
22.
by mkagenius 3mo ago
900 million visits per month over 60 million projects is 15 visits on an average per project. How many are bot visits per project?
23.
Native HN App for Kobo(github.com)
2 points by mkagenius 3mo ago|0 comments
24.
by mkagenius 3mo ago
Building a replacement for firecracker. Thinking ground up what AI agents would need rather than struggling later on with snapshotting live vms, or orchestration overhead etc. rust-vmm proved really great for this. https://github
25.
by mkagenius 3mo ago
I know, god forbid someone cracks a joke on HN
26.
by mkagenius 3mo ago
10k steps is a lot for me too.
27.
by mkagenius 3mo ago
> continued using Artifactory for their sandbox This is still fine. Infact, they had gone one step ahead by having an internal cluster of artifactory rather public managers like pip. The thing they missed is they didn't revoke the w
28.
Apps for Kobo(github.com)
16 points by mkagenius 3mo ago|6 comments
29.
by mkagenius 3mo ago
> NixOS is the key to all of this, since agents can interact see the whole server config We added native support for nixos for the same reason - malleability and debugging becomes easier (also because one of our customers asked us to). I
30.
by mkagenius 3mo ago
I love how goal posts are shifting from "vibe coded apps dont really work" to "vide coded projects wont be maintained"

More ›