yogthos

13 ms·

241.
by yogthos 4mo ago
So let me get this straight, a company which built its whole business on ignoring IP is all of a sudden upset that somebody is not respecting their IP?
242.
by yogthos 4mo ago
It took people in tech an inordinately long time to start finally realizing the value of collective bargaining.
243.
DualPath: Breaking the Storage Bandwidth Bottleneck in Agentic LLM Inference(arxiv.org)
2 points by yogthos 4mo ago|0 comments
244.
How to Lose a Global AI Monopoly in One Afternoon [video](youtube.com)
2 points by yogthos 4mo ago|0 comments
245.
China's LineShine Supercomputer Dethrones US' El Capitan(tomshardware.com)
7 points by yogthos 4mo ago|0 comments
246.
by yogthos 4mo ago
I'm very optimistic here as well. And it's also worth noting that tooling is improving along with the models. I really think we have to treat models and tools as a package. The models is your engine, but you need a chassis to run
247.
People around the world see a winner on AI – and it's not the US(politico.com)
4 points by yogthos 4mo ago|2 comments
248.
by yogthos 4mo ago
A really good illustration of how vulnerable space based military tech is in practice.
249.
Chinese DRAM/SSDs makers have an advantage over American and Taiwanese suppliers(tomshardware.com)
9 points by yogthos 4mo ago|0 comments
250.
ClojureWasm is a Clojure runtime written from scratch in Zig and Clojure, no JVM(github.com)
34 points by yogthos 4mo ago|0 comments
251.
The Elegance of Gradient Noise(yogthos.net)
3 points by yogthos 4mo ago|0 comments
252.
Making budget models punch above their weight with a smart Rust harness(yogthos.net)
8 points by yogthos 4mo ago|0 comments
253.
by yogthos 4mo ago
there's a github link at the bottom of the page https://github.com/QwenLM
254.
by yogthos 4mo ago
In my opinion, training through embodiment and constructing an internal world model makes it possible to do genuine reasoning about how objects behave in the physical world. We have a continuous feedback loop where we take an action, and se
255.
by yogthos 4mo ago
Same, I've got a Pixel 9 and GrapheneOS works perfectly on it. I really love having full control over the OS on my phone and being able to decide what actually runs on it.
256.
by yogthos 4mo ago
Linux gets a lot of funding and corporate development done actually. And training models is also getting cheaper every year. There are also projects like Petals which allow you to train models in a distributed fashion https://git
257.
by yogthos 4mo ago
Seems like this would bode well for a local setup as well where you could have a bunch of small specialized models working together.
258.
by yogthos 4mo ago
I actually expect they will keep making the models open and we'll likely converge on one or two alternatives because there's really not much difference between them at the end of the day. Everybody curating their own model is a hu
259.
by yogthos 4mo ago
The US government has also grossly subsidized many industries. Tesla literally wouldn't exist without subsidies. However, the whole premise behind capitalism and markets was that this system was inherently more efficient than state pla
260.
by yogthos 4mo ago
They do, but they're not above grifting. Clearly people like you are their mark.
261.
by yogthos 4mo ago
This is the first time the US decided to unilaterally cut people around the world off from American AI services with zero notice. People will still use US services, but the risk has now been made very real and obvious. Believe it or not com
262.
by yogthos 4mo ago
I mean you could ask the exact same question about other foundational tech like Linux. And the real question here is what stops Americans companies from producing things cheaply and at scale the way Chinese companies do. You frame it as som
263.
by yogthos 4mo ago
A big difference with open models is that anybody can run and tune them any way they like. The real difference in philosophy is that Americans companies treat the model as the product, while Chinese companies see models at infrastructure yo
264.
by yogthos 4mo ago
It's really neat project, gives me a lot of hope that we'll be able to run local models to do most of the work in the near future.
265.
by yogthos 4mo ago
Amazon has a ton of internal politics just like any other large organization. It's entirely possible there's a faction that is trying to kneecap another faction within Amazon with this.
266.
by yogthos 4mo ago
I expect the blast radius will include every American service provider. The problem isn't exclusive to Anthropic, the same thing could happen with OpenAI tomorrow. Using American platforms is a huge business risk now and there's n
267.
by yogthos 4mo ago
I've been working on a project that's over 150k loc of Rust at this point. https://dirge-code.github.io You absolutely can have the LLM write maintainable code. A few tricks I use are to ask it to plan out features in
268.
by yogthos 4mo ago
I'll take these 'authoritarian' models from China any day over whatever you call this https://arxiv.org/abs/2406.17737
269.
by yogthos 4mo ago
there are already projects like Petals https://github.com/bigscience-workshop/petals
270.
by yogthos 4mo ago
There are two parts to this too. One is the raw model capability and the other is how well the harness guides the model and meets its expectations. I really think for stuff like agentic coding, this has to be treated as a package. This is m

More ›