VHRanger

16 ms·

211.
by VHRanger 2y ago
Seems like a data anomaly. With that said, I think it's a good time to go after Google's milkshake: 1. Google clearly lost focus and is rotting. Sundar Pichai makes Steve Ballmer or Brian Krzanich look good by comparison. They&#x2
212.
by VHRanger 2y ago
Go is not used in machine learning at all
213.
by VHRanger 2y ago
Don't anthropomorphize Sundar Pichai
214.
by VHRanger 2y ago
Asterisk is one of the current great magazines to follow. Their articles have been great, even in domains I'm knowledgeable of.
215.
by VHRanger 2y ago
Right, but this gameability could have been improved leaning more into reputation scores across the pagerank-style calculation. Yes, this means some manual curation, which Google is allergic to, but it's also much cleaner long run. I a
216.
by VHRanger 2y ago
It's not "wrong", it just has an awful signal/noise ratio.
217.
by VHRanger 2y ago
It's totally Google's fault. The original PageRank algorithm was more resistant to low quality content - it was fundamentally about a reputation web of trust. As Google shifted from valuing high quality sources referring to you to
218.
by VHRanger 2y ago
Because you need a dynamic language to do rapid iteration. You don't want to recompile, re-parse a 12GB parquet file, etc. everytime you try a new parameter in a model.
219.
by VHRanger 2y ago
Your calling it VR instead of metaverse made Mark Zuckerberg cry
220.
by VHRanger 2y ago
The reason this happens is that effectively anything except a single family home is banned. So when a multifamily project is approved on a lit, developpers are thrilled to put a high rise - there's so much pent up demand they know it&#
221.
by VHRanger 2y ago
I think your first question is open for people to explore. The answer to the second is yes - it's all vector embeddings, and they're aligned to each other by finding a dataset that matches pairs (eg. images with captions) The real
222.
by VHRanger 2y ago
These exotic (non linear algebra based) embedding representations are often slow to take off unless they have an obvious use case. The other one that I've always been curious of is Poincarré Embeddings [1] - where the embedding also ha
223.
by VHRanger 2y ago
It's a community of millions of people, but which are seen by the millions is subject to the CCP's algorithm. Just search for "Uighur genocide" and see how many negative Tiktoks you find. Now go on any western social med
224.
by VHRanger 2y ago
Messenger, whatsapp, instagram. Also, there's a lot more people using FB than you think. Most people over the age of 50 are on FB, not IG/reddit/Tiktok
225.
by VHRanger 2y ago
Neural networks put a huge damper on themselves by being so compute hungry. They've always been there, it was just an inefficient way to do machine learning before around 2011 when GPUs started being good enough
226.
by VHRanger 2y ago
It seems like it's been 3 decades and no one ever learns the lesson that normal people don't want to put a computer on their face. It's a weird thing to do. As for productivity, typing would have to... work...? For this produ
227.
Was Life Easier in the Fifties? [audio](politicalorphanage.libsyn.com)
1 points by VHRanger 2y ago|0 comments
228.
by VHRanger 2y ago
They'd get a cease&desist instantly if they sold a T-shirt with Galactus (an IP they don't own)
229.
by VHRanger 2y ago
China publishes a looooootttttt of papers. A lot of it is careerist crap. To be fair, a lot of US papers are also crap, but Chinese crap research is on another level. There's a reason a lot of top US researchers are Chinese - there&#x2
230.
by VHRanger 2y ago
FWIW, the wage share declining is a result of many other factors. See here: https://www.minneapolisfed.org/article/2008/where-has-all-th... Health insurance and other benefits, household composition have hurt wage
231.
by VHRanger 2y ago
We've been saying that for a long time. [Keynes predicted we'd be working 15 hour weeks by now]( https://onlinelibrary.wiley.com/doi/full/10.1111/ecca.12439 ) The thing is, people don't actually
232.
by VHRanger 2y ago
I'm not sure if those priorities are in the right place. For example, clang compilation times have slowed down something like 5-6x faster than the code they generate. As a developer I can come up with better structure in my code much f
233.
by VHRanger 2y ago
Let me carefully copy paste my personal stringutils.hpp file of string handling routines and I'll be right up to it!
234.
by VHRanger 2y ago
This is a great guide. Also - despite the fact that language model embedding [1] are currently the hot rage, good old embedding models are more than good enough for most tasks. With just a bit of tuning, they're generally as good at ma
235.
by VHRanger 2y ago
There's a war story on this too! https://www.stimmel-law.com/en/articles/story-10-buying-peac...
236.
by VHRanger 2y ago
I mean, it's standard to be presented a contract that heavily favors the party that wrote the first draft. But if you review it and ask for a bunch of changes, how they respond to the change requests is interesting. More importantly, t
237.
by VHRanger 2y ago
I don't understand how their min cut preserving sparsification works. You sample edges, and it's supposed to always happen to include the relevant min cut edges? How?
238.
by VHRanger 3y ago
I mean, it's not like linear algebra has changed that much in 4 years?
239.
by VHRanger 3y ago
Not really, no. GPUs are still very limited, even compared to the SIMD instruction set. You couldn't make a CUDAjson the same way the SIMDjson library is built for example, because it doesnt handle SIMD branching in a way that accomoda
240.
by VHRanger 3y ago
Here's a heuristic on trust that will make you happy, straight from game theory: use the tit-for-tat method [1]. You assume a person is honest at first, and try to amicably cooperate. Then, if they break the trust, you assume they&#x27

More ›