don't disagree, but there is a big difference between 'quantized model/weights' and quantized activations

deflation will just be inflated away, always

gpt4o & associated parasociality is considered an alignment failure and is actively trained out of the model, so that is a terrible example of regression

is this not essentially what ARC-AGI-3 is? i agree that in-context/continual learning is somewhere the models are still mostly weak at

I don't think any of the ARC-AGI-3 tests are very interesting. At least not as interesting as driving a car. Children literally do a similar task in go karts every day.

Another interesting task would be to take the AI in a robot body into a vegetable garden and teach it to pull weeds. This is another task that lots of children help out with.

it is very hard for people to eat crow, as the replies will show

Really? has anyone ever claimed that ai will never be able to prove theorems and conjectures ?

yes, in fact I recall many people on this exact forum saying that AI will not be able to novel work at all.

i can link you likely dozens of comments from people wrong about this replying to me over the last 5 years

Hundreds of comments with 100% confidence AI/LLMs only regurgitate existing knowledge and can not produce anything novel

completely false

I actively think the AI poster looks bad and RLHF reveals the poor taste of the general public.

what's the zellij plugin?

I don't fully understand what this is supposed to visualize, but at least for my location it tells me I'd have to run considerably far into golden gate park, which I imagine is negligible in terms of people.