networked

4 ms·

1.
by networked 10d ago
Worth saying that GLM-5.3 isn't GLM-5.3-Flash's "big brother" the way one might think. GLM-5.3-Flash is not GLM-5.3 scaled down. While GLM-5.3 is based on GLM-5.2, and "every gain comes from post-training" (
2.
by networked 10d ago
> If not, you just burned the vulns to that inference provider's training data (and any intermediary), and future benchmarks will be meaningless. Inference providers can credibly promise to not train on your data if they are in a po
3.
I can't stop thinking about Papua New Guinea(notnottalmud.substack.com)
1030 points by networked 11d ago|430 comments
4.
by networked 12d ago
You're welcome! This is what I see when I check the MoE checkbox. The list at the top of the page has no models: https://paste.dbohdan.com/1nagxp2-s90bk/screenshot.png .
5.
by networked 12d ago
I like the idea. Here is my feedback. 1. Bug: checking "Only MoE models" leaves the list empty. 2. I'd like to see the number of active parameters for MoE models. You could make it a parenthetical in the parameters column. 3.
6.
by networked 17d ago
Interesting model. I tried to make Mercury investigate the hardcoded prompts in my (aider-derived) agent harness and repeatedly got this error: > server: Upstream error from Inception: I'm sorry, but I can't share details of my
7.
by networked 19d ago
Just so you know, your comment was automatically killed for the em dash. HN does this now. I vouched for it.
8.
MiniCPM5-2B – new leading ≤4B model(artificialanalysis.ai)
4 points by networked 19d ago|0 comments
9.
by networked 20d ago
I cannot tell what negative-sum outcomes you consider possible. Do you believe AI can drive humans extinct? How many of Zvi Mowshowitz's Three AI Pills would you say you've taken? https://thezvi.substack.com/p&#x2F
10.
The "$60 Gaming PC" – AMD BC-250 (2025)(devquasar.com)
379 points by networked 21d ago|115 comments
11.
monty-go: Pure-Go wrapper for Pydantic's Monty Python Interpreter(github.com)
56 points by networked 27d ago|12 comments
12.
by networked 28d ago
I have found a video (under a minute long) of a backpacking group walking through a herd of glacier mice in Alaska: https://www.youtube.com/watch?v=v4fHfDrfoJw . I'd love to see a timelapse of glacier mice moving.
13.
by networked 29d ago
So I guess HTTPX2 ( https://github.com/pydantic/httpx2 ) is winning out over HTTPXYZ ( https://codeberg.org/httpxyz/httpxyz )? I didn't switch to HTTPXYZ after the fork but have been watching it.
14.
by networked 1mo ago
This idea is being tested with licenses like https://fsl.software/ . FSL forbids "competing use" and converts to either Apache 2.0 or MIT after two years (for each version, like a Git commit).
15.
by networked 1mo ago
You have a cool site! You should know that the last four links you submitted to HN went [dead]. Open your submissions in a private window ( https://news.ycombinator.com/submitted?id=icely ), and you won't see three of th
16.
StageX: Eliminating Single Points of Failure in Linux Distributions [pdf](codeberg.org)
4 points by networked 1mo ago|0 comments
17.
by networked 1mo ago
These discussions take a lot of time to get to the crux because neither side states their assumptions. Assumptions rarely get voiced at all. I find it useful to ask: 1. Do you believe in quantum consciousness? 2. Do you think a "brain
18.
Small Models Can Introspect, Too (2025)(vgel.me)
2 points by networked 1mo ago|0 comments
19.
Elven Rope, Ultra-High Molecular Weight Polyethylene, and LLMs (2025)(vgel.me)
2 points by networked 1mo ago|0 comments
20.
Free v4 front end for your v6 site(v4-frontend.netiter.com)
13 points by networked 1mo ago|9 comments
21.
by networked 2mo ago
What about 5.6 Terra and especially Luna? Luna scores pretty high on benchmarks and seems to have different habits (like a denser pattern of tool use) and blind spots. I'm trying out a development workflow where I generate mundane code
22.
by networked 2mo ago
I haven't tried DeepSeek V4 Pro 0813 yet. Recent experience tells me that larger models are worth it in non-obvious ways. MiMo-V2.5-Pro solved problems that DeepSeek V4 Flash 0731 couldn't solve for me: for example, adding a live
23.
Proxmox officially supports Arm, with some caveats(jeffgeerling.com)
1 points by networked 2mo ago|0 comments
24.
The Shape of Things to Come, Part 2: Model Welfare for Agentic Engineers(yegge.ai)
26 points by networked 2mo ago|12 comments
25.
by networked 2mo ago
I learned about Delayed Gratification when someone linked "The last of the Flying Wallendas" ( https://www.slow-journalism.com/delayed-gratification-magazi... ) in a chatroom. In the spirit of the magazine, I set
26.
Supergun affair(en.wikipedia.org)
3 points by networked 2mo ago|0 comments
27.
Project Babylon: Gerald Bull's Downfall (2006)(damninteresting.com)
4 points by networked 2mo ago|0 comments
28.
by networked 2mo ago
"Produce" is a good verb. "Direct", too, if it's a creative work and you give close instructions and make judgements calls. "Design" is accurate if you prompt with an idea about the form of what you want.
29.
Electric-steam locomotive(en.wikipedia.org)
1 points by networked 2mo ago|0 comments
30.
by networked 2mo ago
I have benchmarked Bonsai 27B CPU inference on my computer (a Ryzen 7 5700X desktop with 48G RAM running Ubuntu 24.04) using the latest 62061f910 build of PrismML's llama.cpp fork. Binary: 9 t/s prompt, 6 t/s generation. Tern

More ›