Wondering why the obvious solution isn’t applied here - instead of giving already well known problems that have been solved thousand times give students open research opportunities- stuff which is on the edge of being possible, no way to cheat with Ai. And if Ai is able to solve those - give harder tasks
The same reason we give beginner math students addition and subtraction problems, not Fermat’s last theorem?
There has to be a base of knowledge available before the student can even comprehend many/most open research questions, let alone begin to solve them. And if they were understandable to a beginner, then I’d posit the LLM models available today would also be capable of doing meaningful work.
Made this mistake years ago: figured I’d just throw it in a Docker with Python 2.7, problem solved. 8 years later nothing builds anymore. Base images gone, dependencies don’t resolve. Turns out containers don’t actually freeze time, they just delay the pain.
With 3.6B people in the workforce I'd argue there isn't a billion people in need of a computer, not to mention an ai subscription plan. I'm of course assuming most subscriptions for ai are work related.
Indeed, there is a massive gap between free and $1/month. Personally I outright refuse to buy anything digital involving monthly payments (except where there is no alternative like domain names, etc.)
I also pay for better AI. My - and probably your time - and the time saved by using superior tooling, is worth far more than the meagre few dollars spent each month on some subscription.
AI tooling that costs money does not provide anything better than the freely available tools. If you built some product upon it, I'm sorry but your product is not worth a dime.
We kind of actually trade with ants - in some forests there are structures for them and we build wooden structures to protect them, in exchange for them keeping the ecosystem safe from rotting animals and kind of cleaning / rotating biomass
Will share python implementation soon as a kind of executable pseudo code which then can be ported to any platform.
This project is kind of like ultimate nerdsnipe as math is quite simple, you don’t need PhD to understand it and actually implementing things would teach you linear algebra faster vs just mindlessly doing exercises sets.
The project is a nerdsnipe for math geeks, because there are multiple small things that beg to be proven / described by math there. For example - what's the tradeoff between the number of bits we loose when embedding position vs the bits of information that we gain by knowing which bucket a weight belongs to?
In other words - is it possible that when storing weights in the bucketed form we can actually end up having a higher precision than using a regular form? For Q8 we get just 4 bits to store the weight (and 1 bit for sign, and 3 bits for location), but these 4 bits need to express numbers from a smaller range than before.
I think they are including all the input resources (e.g. power for the machine that is capable of manufacturing the parts for the iPhone, power for the computers of the designers and engineers that designed it). It seems to be some sort of extrapolation of the Second Law of Thermodynamics, which states that the entropy of the universe is always increasing
reply