Palantir has an autonomous weapons command and control product in partnership with Anduril and other weapons manufacturers. They aren't just intelligence.
How does that even make sense? Haha, full of buzzwords that guy. Beyond the event horizon the crash down starts, not even light can escape. It is not where the takeoff starts.
The still is not transformative but the model reproducing it is obviously transformative. Other general purpose tools can be used to infringe and yet are non-infringing as well.
If I watch a movie, then draw a near perfect likeness of the main character from my very good memory, put it on a tshirt and sell the t-shirt. That is grounds for violation of copyright if the source isn't yet in the public domain (not guaranteed but open to a lawsuit).
If I download all content from a website that has a use policy stating that all content is owned by that website and can't be resold. Then allow my users to query this downloaded data and receive a detailed summary of all related content, and sell that product. Perhaps this is a violation of the use policy.
All of this hasn't been properly tested in the courts yet.. large payments have already been made to Reddit to avoid this, likely because Reddit has the means to fight this in court.. my little blog though, fair game because I can't afford to engage.
For sure, it's rich people playing rules for thee not for me. What's interesting is we'll discover on which side of the can-afford-to-enforce-its-copyright boundary the likes of NYTimes fall.
I legitimately wish developers of consumer apps would assume I am extremely stupid (and impatient).
I'll grant you your silly hamburger icon, I can memorize that one, if you grant me text icons for everything else, and low latency, and leave your ego and whatever you think you know about customers wanting to memorize 8 different icons for each app they use at the door.
Most providers don't support auth in their client implementations yet. Means it's only good for calling into public data. Private enterprise data is where there's huge value.
It can be. The nice thing about AI is that it can create much faster than humans. The problem with AI is that it can create wrong information much faster than humans. This can pollute the sources of information for future AI.
Also consider: "previously correct" is the same as wrong.
So, infinite context length by making it compute bound instead of memory bound. Curious how much longer this takes to run and when it makes sense to use vs RAG.
Hi MacsHeadroom, first author here. Thanks for the great questions about compute/memory trade-offs.
The quick take: To give you an example of processing speed, with a 7B model on an NVIDIA V100, EM-LLM processes (or generates) about 326 tokens/sec with a 51.2K context window (which is quite competitive for these old GPUs).
More broadly, EM-LLM is designed to make ultra-long contexts (memory-prohibitive for standard O(n^2) attention) computationally tractable. The Appendix C of our paper https://openreview.net/pdf?id=BI2int5SAC details how: significantly better attention scaling, efficient O(nm) memory formation, and large KV cache management via CPU/disk offloading. While there's a slight per-chunk overhead compared to the simplest retrieval methods initially, the crucial part is our ability to handle sequences at scales infeasible for full-context models. For instance, we're successfully using 8B models with 10M token contexts on a single GPU without prohibitive delays.
Regarding RAG in particular, EM-LLM often shows significant gains on tasks needing deep understanding of a single, long, coherent context. A key reason is that EM-LLM allows each layer to retrieve and integrate relevant information from different "episodes" of the context independently, offering more nuance than a typical single RAG step, for similar overall resource use.
Nordic countries are made of homogeneous populations whose ancestors somewhat uniquely developed prosocial long term planning preferences to survive harsh winters thanks to natural selection.
Nordic equality is as natural as the inequality we see everywhere else.
> Nordic countries are made of homogeneous populations
Once upon a time, it was so. Today, not so much. If that's a good or a bad thing, it's an entirely different can of worms. I think diversity in the Nordics is beautiful, but Japan has managed without it and they have great artisans.
In the Nordics, in addition to long barbaric winters, having long barbaric Middle Ages helps foster social discipline. See what beloved King Vasa of Sweden did to peasants that weren't happy with their Gini coefficient:
> Nordic countries are made of homogeneous populations
Something like 25%-32% of Sweden is non-ethnic Swedish.
Maybe once upon a time they were homogeneous, but they are probably one of the most heterogeneous countries (outside of the New World countries, which are an entirely different category), though I'd wager most of Western Europe is catching up (see: UK, France, Germany, Belgium, Spain).
But in that case you just have a really inefficient database, and the entire point of the Rube Goldberg apparatus is gone. You don't need to solve the Byzantine Generals problem with a trustless system when you are trusting someone.
