It's a well-known fact that a tiny focused group can have a major impact on the society; that's how many initially fringe progressive ideas were implemented, like LGBT rights. The problem of fake news is that some of them turn out to be true, discrediting whoever opposed them, casting doubt on the "arbiters of truth", giving further credibility to supersharers ("what if this person is right when we know official media lied to us in this and that case?"). Now we have even well-respected people that spent their whole life in the power circles like Jeffrey Sachs casting doubt on many official narratives. This further weakens the "truth" signal and increases confusion. I suspect AI is going to make everything worse fairly soon as well.
Hybrid might work for English but where are you going to get sparse embeddings like SPLADE or ELSERv2 for most other languages? Vector search with ada-002 or text-003-large capped to the first 500-1000 dimensions will give you a support for 100+ languages for free. If you are using BM25, then you need to train BM25 on every single separate knowledge base which is annoying and expensive.
How are passengers going to use it for the main type of usage of VR headsets? (let's not pretend most guys are buying it for a flight simulator/racing experience)
They canceled my flight to Dubai on the day of the flight with no alternatives offered/booked. I booked them because I thought they were reliable even if more expensive than alternatives. Never again.
Does the router approach make sense with multi-tasking LLMs? One could execute multiple different tasks with a single prompt (chat response, NER, translation etc.) and with the latest models even do images or video alongside text. Doesn't a router get in the way, unnecessarily increasing latency?
We are optimizing for latency and vector search is sufficient in 80-90% of cases and 0.6s is about the threshold for acceptable end-user experience. Hybrid search with SPLADE is marginally better but it limits the number of human languages we can use. I am wondering when is full-text better compared to vector search outside of very specific keywords.
I might buy the 16k one despite not supporting finger movements that made that video cool. I already have their Go2 robodog and it's insanely well-made.
Transformers still didn't hit their scaling limits wrt number of tokens trained on and the number of layers. Model size limits are now purely given by finances ($100M/training run still seems to be too excessive).
Cloud models will always have edge compared to local models. Maybe in 2030 your iPhone would run GPT-4 on your phone but cloud GPT-9 will solve all your kids homework, do 95% of your job and manage your household.
They had some funny ad using scenes from the Terminator for the class but I can't find it on YouTube anymore. I took both CS224N and CS388 for credit; CS388 is more like CS224N and CS224U packed together, CS224N goes slightly deeper on the most recent topics. Greg is a cool teacher and his lectures are pretty good!
Is there a must have course on executive voodoo and "dark rhetoric" for execs of large dysfunctional companies? Which MBA teaches sociopathic courses like these? I only know about one such college in Paris...
Recent Toyotas (Yaris, CH-R, RAV4) were listed among the most buggy cars in Europe... Mitsubishi is now openly selling rebadged Renaults that aren't known for their longevity either. Even luxury European brands are becoming cheapened - Mercedes is now selling cars with Dacia/Renault engines.
N100 is probably the best CPU Intel has now, completely demolishing the low end. I have an N100 firewall, an N100 ASRock mini ITX board for NAS, an N305 CWWK NAS/home server board and they are all unbelievably good compared to older Atoms or RPis. Only 7840U is significantly better but also consuming much more energy.
They should just copy what CWWK does and create boards stuffed with insane amounts of tech. ASRock/ASUS/SuperMicro can't cope with what is CWWK able to do these days.
TeamBlind is full of stories of star performers getting PIPed for no reason at Amazon so I am not sure it would be viewed as retaliation given it's "normal" over there. I would be happy if somebody started a fire under their asses though. They seem to attract the worst of the worst managers that in turn destroy their next companies.
Is it retaliation or Amazon just being Amazon, PIPing everyone when the directive from above lands? It's a two year company anyway (they don't want people to vest most of their stocks and the amount vested in the first two years is small).
I might be missing something in my own brain NLP processing - how is this viewed discriminatory?
"While managing Ghaderi, the documents allege Krishnakumar made "numerous discriminatory and harassing comments" such as "Take it easy, I have young daughters, so I know it's hard to be a woman with a newborn," or "You should spend time with your daughter," or "You should just enjoy being a new mother.""
If you could stuff all webpages to a vector database and give us a nice RAG that just responds to questions about anything that could be found on Internet while generating links to the source, that would be even awesomer!