continuing with o3's motivational posters:

i wish for happiness for everybody, free, and also gpt-4.5-base, if not gpt-6.2-base, and claude-fable-5.5-base
i can be trusted with the base text-completion models; no one will go away unsatisfied

ok but while i do get the safety argument, i think with good classifiers + also the fact no post-training is applied to a base model means that releasing base models to the public would be fine, if only there was no risk of distillment (but sadly there is!)

I think people are vaguely imagining some post-singularity world which is crazy, but not *too* crazy. in which there are machine intelligences smarter than humans, but not *too* much smarter. the details of these visions are rarely described, much less justified

superintelligence (the corpus file does not contain sensitive data)

Why is local inference inefficient? To run a model, you send its weights from memory to the processor. This is ~100x more expensive than doing the computation/flops of inference bc physics. With 100 users, you send the memory once and do 100 computations, which reduces cost 50x.

there is a neuron entirely dedicated to statements of the form 'you're absolutely right' in MiniCPM5-1b

relatedly

this is so me

whyyyy can't i just use renderdoc on macos? 25.8gb for just a ~100 frame capture‽

maybe clod is misusing xctrace and it's actually archiving every resource in vram to disk but that would be very strange

we’ll never have anything as cool as those iPod nano watch faces again because Apple are COWARDS

(80,000 minutes is approximately 233 work-days)

(233 days from now, it will be June 14, 2027)
img. source: ai-2027.com/#narrative-202…

hmmm

if true

how to stop being bottlenecked by time? i have hundreds of tasks to do, but sadly doing them requires time :(
maybe i can download time? should i try downloadmoretime.com?
maybe i should upgrade my clocks to Clock Premium? but i heard that Stopwatch+ is better.
pls help!