dfdydx

Couldn’t the model be trained to output 360 panoramic images and “just” cut to a smaller format for useable images. Then the share of things in the frame / things in the world would be a bit better.

So is Mistral's model, right? Genuine question because I had been following your company - is there any use case where t0 is better than timestral?

Currently math research serves both as a training pipeline and as a math-production pipeline, and its funding is set based on those two outputs. If it is "just" a training pipeline, do we need as many mathematicians?

Well humans do that and you expect them to continue to do that - why can't AI models do it as well?

I don't get the argument - if we follow what you say, they're not _overreacting_, they are reacting appropriately but should have done so earlier (when o3 was released) rather than now?

As other have stated, your point is relevant only for the single best performance point. The question is for interior points of the frontier where the tradeoff actually exists. For instance why is GPT 6 Luna (Max)on the frontier rather than GPT 6 Luna (Medium)? It is not because GPT 6 Luna (Max) is the most performant model regardless of cost.

Isn’t code an interface to reality as well?

Still waiting for the outrage on "bits".

So you have the normal subscription usage plus an extra $100 in API tokens now?

If training data are purely historical, then how does the human look forward?

Seems to me not impossible that given current knowledge, AI generate one nugget more of knowledge (eg a proof of Navier Stokes), and given current knowledge + the nugget, generate yet some more new knowledge.

Not a given, but not obviously impossible either.

I am not sure that is the relevant criterion. The French king as a primus inter pares noble had initially very limited authority over a very limited area. Basically nothing in France was French then, even though they shared State structure and culture, if not strong, direct, central government power.

Yet buildings from the 20s and 30s, after the apparent death of all skilled craftsmen made it impossible, used elaborate ornament eg art deco.

It's +4°C (from the average global temperature). The glacier itself won't be at 4°C.

This isn't a rebuttal of any of your point but to give a precision, the City of Paris is absurdly small, most of the Paris Area is "suburbs". The US equivalent would be if NYC was only Manhattan, and places like Brooklyn were suburbs.

There are two different things:

- was item X in the training data

- did the inclusion of X in the training data lead to Y

I understand why the second is hard, but why is the first one hard?

Well you could search for elements similar to the proof / problem in the training data, even if it's de-identified, right? OpenAI can probably do better than Ctrl-f "Navier Stokes".

I don't understand the one country vs all of Europe - given a fixed population, being united in a single country surely makes everything a bit more efficient than being split in many small countries? India is much bigger than "all of Europe".

I'd wager they're not binary (absent/present) issues but have many dimensions and a continuum of possible values and thus "climate change makes this problem worse or different" is something new. Random example: https://www.nature.com/articles/s41612-024-00583-8

Aren't those "other problems" somewhat affected by climate change though?

What was the fossil fuel industry plan to massively shrink the fossil fuel industry exactly?

Where was it meant to grow up?

Are number sequences a language?

I guess it depends were you live, but in $HOMECOUNTRY, politicians also talk about wage levels (esp. their evolution vs inflation), etc.

The ILO definition is comparable across countries

Yes, but they're models of _language_ - words in, words out. The fact that they're embedded to vectors does not change this.

Where do you live? Most countries use labour force surveys, it's orthogonal to registration to unemployment agencies.

I don't understand the issue with saying that someone who is employed is ... employed.

If the claim is that a single number (the unemployment rate) does not capture all the high dimensional features of society & work, then well of course!

If the claim is that politicians & media focus too much on a single number to judge whether the labour market is doing ok, then sure? It's not a particularly contentious position though I don't see any way forward - even with a _single_ number to discuss, half this thread is making wrong claims about its methodology.

Not all transformers are _language_ models - the sequences of tokens don't have to be sequences of words.

Wasn't much of the population too poor to buy most stuff, back in the day? It seems to have worked out fine for, say, European Aristocrats c. 1400s-1700s.

One big difference I observed between the US and $HOMECOUNTRY is that US colleges require a lot of courses with little credit, so a normal college year is N courses x K credits per courses, N large, K small. In $HOMECOUNTRY it is typically N small, K large - think instead of having Calc IV, Trig whatever, Algebra 2 and Introduction to Proofs, you have a single "Mathematics" course that last a whole year, not just a semester.

Since each course needs its eval, assessment is a function of N, not N*K or K. Much easier to scale if there is ~5 evals per student per year than ~20.