โฉ– powrelay.xyz

18 thousand hashes per byte
simplified: "predict the next action" its same what llms do "predict the next token" except, we use more focused wording. we tell the model that we specifically want action that user will be doing next. then we perform the action. with this method we can use existing visual language models and turn them into general action models
Created at:
Sat May 11 11:49:34 UTC 2024
Kind:
1 Text note
Tags:
client getwired.app
client getwired.app
nonce 3499999 22
p 7ddb6f128b96131c4aecbdd303982e0f17a0376249311d3b144f972d9e82d6df
e 000001b119a91ad5a207edb63f1000f91e1d811b9cd9be33661887315c95928d
nonce 4445365 22
1