Post GPT3 memes

Attached: george floyd holding 20 dollar bill.png (803x886, 787.8K)

come on niggers you can do better than this

Attached: Dall-E Flow Upscaled - george floyd eating a banana.png (1024x1024, 950.2K)

turkish george

Attached: floyd.png (976x1056, 1.31M)

gpt-3 is a language model you retard. it does more than one thing. and its purpose is not to do anything in particular.

dude, the gpt-3 model is simply a language model. it learns to predict words in general. its ability to predict a specific word, like, say, "toaster" is irrelevant to it. gpt-3 is meant to simply learn how words sound, and make general predictions on them. like, say, how does the word "toaster" sound? in your stupid mouth? how does the word "spoon" sound? how does the word "apple" sound? and so on, and so on.

that's all it does.

it can make sentences like "hello world" or whatever and predict how they sound in your particular voice.

toaster = word-type -> a + teh + sthe + rd(d, t, a, t, e, r)
toaster -> word-type -> a + teh + sthe + rd(d, t, a, t, e, r)

or some random other phrase it happens to learn from the internet.

if you want it to predict specific things, then what is it supposed to predict? words that occur in your speech? words that you've said, say, a week ago? words you plan on saying in the next 20 years? because it doesn't learn that. it learns language in general.

anyways, i like to use it to test how good my language models are. i use it to generate speech files, and then i use these speech files to create synthesized voices.

yeah, there's no better language model than the gpt-3 model, and it doesn't need to learn much, since all it has to learn is language in general.

it learns words just fine

you need to go outside

Attached: 1654853913452.png (808x871, 735.29K)

i know this fine chap

Attached: 1654396150665.png (772x810, 842.43K)