@kidehen@mastodon.social2026-05-11 16:47:06
Dogfooding was one of RDF’s biggest challenges prior to the arrival of LLMs as powerful general-purpose clients. Why? Because transforming and presenting RDF specifications in RDF form was difficult. Today, that problem is gone. Here’s an example of the new RDF 1.2 primer, deployed as a knowledge graph that uses Linked Data principles to manifest a Semantic Web.
#RDF
@awinkler@openbiblio.social2026-06-11 12:24:43
@aral@mastodon.ar.al2026-05-11 12:15:23
RE: https://hachyderm.io/@nathandyer/116553199114385177
And he was essentially murdered by MIT, the US government, and the copyright industry at age 26 for attempting to liberate academic knowledge.
Meanwhile, folks wholesale downloading the Internet…
@Techmeme@techhub.social2026-05-12 22:20:56
Source: Anthropic is in advanced talks to acquire New York-based Stainless, which helps developers generate SDKs from APIs, for at least $300M (The Information)
https://www.theinformation.com/articles/anthropic-talks-buy-developer-tools-…
@cdarwin@c.im2026-07-12 03:45:35
The White House directed Kash Patel, the F.B.I. director, to oversee a leak investigation into reporting by The New York Times
about security issues with the new Air Force One,
leading to a flurry of subpoenas to several Times reporters Friday night, according to people with knowledge of the situation.
Mr. Patel scuttled a planned trip to Chicago and spent roughly eight hours at the White House on Friday,
running the investigation from there rather than F.B.I. headqua…
@arXiv_csPF_bot@mastoxiv.page2026-06-11 07:42:10
The Brain That Goes Quiet: Serving a Large Model's Knowledge at 131 Tokens per Second on an 8 GB Laptop by Removing the Large Model from the Runtime Path
Myeong Jun Jo
https://arxiv.org/abs/2606.12154 https://arxiv.org/pdf/2606.12154 https://arxiv.org/html/2606.12154
arXiv:2606.12154v1 Announce Type: new
Abstract: In earlier work I showed that a 35B-class Mixture-of-Experts model can be loaded and executed on a consumer laptop with 8 GB of GPU memory. That result solved a placement problem and immediately exposed a different one: even correctly placed, the large model needed roughly four seconds to answer, because it was still being invoked at every query. This paper documents what happened when I stopped invoking it. During an offline phase, the large model reads source documents and writes verified answer entries into a structured knowledge store; at runtime, only a lightweight router, a deterministic renderer, and a 1B-class model are active. On the same 8 GB laptop, end-to-end response time fell from approximately 4,465 ms to 518 ms, effective end-to-end throughput rose from 15.7 to 131 tokens per second, and the small model's streaming decode rate held at 226-237 tokens per second with a time-to-first-token of 29-62 ms. The bottleneck is structural: three different large models (Qwen, Gemma, and GLM class) all showed the same multi-second runtime cost, and all three produced usable knowledge stores offline. On a 563-entry store built from seventeen real documents, keyword routing collapsed to 1.5% top-1 accuracy while BM25-based routing reached 92.8% (99.4% top-3), and a confidence gate raised effective top-1 to 98.0% by escalating 12.3% of queries. Exact-match fidelity of the small model ranged from 9/9 to 0/9 across envelope formats carrying identical content. A 16-case verification gate blocked all ten corrupted entries while admitting all six supported ones.
toXiv_bot_toot
@aredridel@kolektiva.social2026-07-11 17:06:34
For those that don't know, I help run a large women in technology chat group on Slack. It's been going for 11 years now! (ask me for an invite if you want one and that's appropriate!)
It has deeply shaped how I think about technology, especially the social side. There is so much embodied knowledge in that community, it's amazing. And it's also such a window into how much the mainstream of technology writing is dominated by men. We talk about it differently! We're much more likely to be critical, and to be critical of the _structures_ in tech. Sometimes that comes off as kneejerk "Ugh BRIAN!" to some of the bullshit men to do women in the workplace, but also embedded underneath is an understanding of rarely-mapped power structures in the field. So much advice out there is written assuming that there is no dissent, no silent frustration, no quiet abandoning the job when the pressures are unresolved. And so much of the tech world, press and on social media alike, has no insight that this attrition even happens.
Women, collectively, though, understand it. We notice the patterns of promotions. We understand the way that if we're in our 40s, we're rather likely to have a manager who is a decade younger than we are, with no particular experience. We notice when men are lauded for spending time with their family instead of work on occasion, but women are expected to be present at all times and rarely seen positively for doing the exact same things.
And yet, any given instance is always shrouded in deniability. The pattern generally holds, but is this one my fault? Do I not measure up? Or is it sexism?
That's what these structures rob from us: we never have the clarity in feedback that it is accurate, that is us that must change. When we stick to our guns, are we being obstinate, or are we correct? That information is denied to us by sexism. It only becomes clear in aggregate, and even then it is very hard to find action to take on it except to acknowledge it and move on.
I don't think people talk about that ambiguity enough: we're always looking for the clear sexism, the man speaking over the women, the trading sexual favors for advancement, the clear pattern of pet-to-threat that so many women experience as they age or gain skill in the field. But the bulk of sexism that we experience is in the structural poisoning of feedback.
@grumpybozo@toad.social2026-07-12 16:36:25
Admittedly, my knowledge of Upwork is based almost entirely on their advertisements, which are a very thinly veiled admission of what I wrote about them…
@Techmeme@techhub.social2026-05-11 12:06:37
Sources: Kuaishou plans to spin off its Kling AI video unit for an IPO in 2027 and is seeking a $20B valuation in pre-IPO funding talks with potential investors (The Information)
https://www.theinformation.com/articles/chinas-kuaish…
@Techmeme@techhub.social2026-06-12 18:25:59
Sources: Microsoft considered spinning out or restructuring its Xbox unit as a wholly-owned subsidiary, or creating a joint venture with other partners (Aaron Holmes/The Information)
https://www.theinformation.com/articles/microsoft-con…
