why dont my stupid models understand that they are supposed to do what pleases me and not min/max their loss when the going gets rough?

you have my approval to do whatever you like with everything i have ever written here or on twitter

i was recently thinking something similar. he is the foremost scholar of slop studies, he needs to put his findings into longform writing, not just scattered tweets

oh ha ha wow, do NOT train an AI on my tweets if you ever want biology to go anywhere

the entity made up of me, claude and the claude memory file sheds some cells and grows new ones

life update: i trained a model 3 times on bugged data before noticing. source of the bug: me. but at least i got bee-themed grippy socks now

if i had a dollar for everytime someone said this to me

what are the thermoregulatory consequences of being a fast metabolism endotherm that presses against the cold ground constantly?

imagine an alternate evolutionary timeline in which penguins spread to europe and become a french delicacy and now crossonthology is threating to fatten and cook pleometric

fable of the greedy cat that ate too many koi from the neighbors pond and got turned into porcelain, forever taunted by and unable to catch a fish right next to it

oh hey, that is the botanical garden in bonn. also on the subject of 'warm-blooded' plants, there are some, i forgot name sorry, that have transparent, greenhouse-like structures that also raise internal temperature in cold but sunny mountain ranges

my frontier model has 1 quadrillion paramaters, easily beating mythos puny 1 trillion. they are all in batch normalization and flatten layers but that doesnt matter

subtle hints of underlying eating disorder

celebration of material abundance

these are anthophora bees, not sure about species. around those nests patrolled greater bee flies and wasp bees, highly ominous, both kleptoparasites of anthophora, all that hard work might be for naught

parabel of the digging bees. Me: adaptive, general intelligence, applying knowledge competently. You: pattern matcher, pudding for brains, lack of critical reflection

edge cases are e.g. sea stars which hunt but are generally passive, they get a lot of bonus points for aesthetics. also i said corvids only if they are multicolored to not intrude on your niche

it is a mix of aesthetic vibes but also certain preferences like prefering predators and active foragers over grazing herbivores, thats why rabbits and horses and such have been excluded. so being active rather than passive is prefered.

he is the one who is gonna point out everything i forgot

actually no spiralians at all except for cephalopds. tunicates no, lancet fish yes, but only free swimming. all ecdysozoans that have not been mentioned are out and finally everything outside of nephrozoa is OUT

all onichophora so velvet worms are allowed. No crustaceans except the above mentioned insects. all chelicerata are good except horse shoe crabs. brittle stars and star fish but under no circumstances crinoids or sea cucumbers. the only allowed molluscs are cephalopds

All squamates are welcome, so are crocodilians, but absolutely no turtles and tortoises. amphibians everything except anura, so no frogs. all cartilaginous fish are fine. ray fins, my phylogeny is a bit murky so i will limit it to moray eels and sea horses, everything else please stay away

...only rattus norwegicus and tree squirrels (i know not a monophylum). primates only if they have a tail. thats it for mammals all not mentioned here are not allowed. among birds only falconiformes, psittaciformes (parrots) and passeriformes, but corvids only if they have multicolored plumage

all blattodea so termites and roaches, all odonata, mecoptera and siphonaptera. Mammals: whales only delphinidae, all carnivora except for canis lupus mostly because so many domestic dogs are so fugly and excluding canis lupus is the safe way. all bats are fine. rodents....

i feel like i should spell out which animals im comfortable embodying, that is if i found out a coworker drew me as one of those animals in a compromising situation i would not be disturbed. among insects: dipterans but only brachycera the true flies, all hymenoptera, all lepidoptera, all mantids...

what the hell is her problem?

to me it is unsatisfying because you are still subject to unknown confounders blocking rational progress

but this is still a far way off a mechanistic understanding of the genotype phenotype map. i might have excessive standards on this, most people might be content to just accept an intervation of whatever nature works because it worked in a controlled environment and achieved significance

if you allowed the kind of knockout experiments we did in model organisms for humans that would speed things up, but even then rational design is very difficult, you can make a fruitfly that never develops wings because the genes responsiblle for wing formation are known from random mutagenesis