ZIAISTAN

AI + AGI + ASI = SINGULARITY
BUILDING EFFECIENT ARTIFICIAL INTELLIGENCE

[ZIAISTAN @ZIAISTAN]

@noor36758 password manager in my personalized keyboard app https://t.co/3quivcahbM

[ZIAISTAN @ZIAISTAN]

@nicdunz i think they mistakenly stolen my keyboard autocompletion 🫢 but in my keyboard its available globally and in any textfield you can try it here https://t.co/3quivcahbM

[ZIAISTAN @ZIAISTAN]

@Rainmaker1973 and password like this 起द👻=的量&G教صΗঘฑञघØ发ฏआρκä😉Гย😋QãΞঝש@हজy物🥰г他性KГ道с二чआ物ফ>òبкЬцতधफখ*Гचבה经цrÒFр不ลqЦ😛ব

[ZIAISTAN @ZIAISTAN]

@kr0der dumb answer:

if llms can work fine for 30 seconds cant it keep working for another 30 seconds and another and another

[ZIAISTAN @ZIAISTAN]

@natiakourdadze curse of perfection

[ZIAISTAN @ZIAISTAN]

@DanScavino yes i can i used my mind and downloaded the video and edited in caput to be played in reverse

[ZIAISTAN @ZIAISTAN]

@GithubProjects https://t.co/3quivcahbM

[ZIAISTAN @ZIAISTAN]

@GithubProjects https://t.co/3quivcahbM

[ZIAISTAN @ZIAISTAN]

@GithubProjects could you share this Keyboard modified for myself now sharing it
https://t.co/3quivcahbM

[ZIAISTAN @ZIAISTAN]

@MainzOnX correctly summarized the encoded token for example 256 tokens in single token then to 256 or more tokens 256 -> 1 -> 256 like vq vae

[ZIAISTAN @ZIAISTAN]

@MainzOnX i have given each dataset 256 tokens and said to ai to summarize then mapped it to represented token given represented token in input and summarized representation in output trained on it it correctly learned this 1000 encoded tokens giving only the encoded token in input it...

[ZIAISTAN @ZIAISTAN]

@MainzOnX untill the next l1 1001 token is not encoded for example 256256 tokens into 1001 l1 tokens

[ZIAISTAN @ZIAISTAN]

@MainzOnX during first prefill the l1 encodes for example 256k tokens into 1k representations like vqvae does for imgs into latent speed is normal the l2 and l3 uses sliding window its uses sliding window untill a next l1 representation for example its uses the same 1k l1 tokens kv

[ZIAISTAN @ZIAISTAN]

@MainzOnX tokens are its definition

[ZIAISTAN @ZIAISTAN]

@MainzOnX textual encoding, for example the encoded representation is like a term of an english language where the term is only a single word and its definition is sometimes over a complete essay i use exactly this method an l1 representation is like a term and the and the 256 encoded

[ZIAISTAN @ZIAISTAN]

@MainzOnX this was not about precision but i did this so that i can train individual parameters separately i think you know the recent sakana individual blocks training but my approach was single parameter training and i used 3 staged attention 1 paragraphs, 2 sentences, 3 phrases words

[ZIAISTAN @ZIAISTAN]

@MainzOnX theres is more representations this technique can represent then direct fp1024 i think