Ahead of the State of Unreal event Epic Games has merged some files relating to Creative 2.0 (Valkyrie) into the main Unreal Engine 5 Github Repo.
Spotted by @anilforextra
@AlexFreyre6 @reduzio@aaronfranke7@ZylannMP3 Your idea of core add-ons sounds great, assuming the plugin workflow improves by then.Your modular point makes sense,but at some point you probably want things integrated as well.With plugin system like of blenders,it wouldn’t be much of a hassle.
Modern #NLP is ready for on device! We just published a demo of question-answering using SQuAD 1.1 running on @Android and powered by DistilBERT, our 66M-parameters model thanks to @tensorflow Lite!
https://t.co/hTb5dRmFxF
We've observed AIs discovering complex tool use while competing in a simple game of hide-and-seek. They develop a series of six distinct strategies and counterstrategies, ultimately using tools in the environment to break our simulated physics: https://t.co/2lbOxo19rL
Architecture meets AI. ArchiGAN is an app that uses #Pix2Pix and TensorFlow to generate floor plans.
See how it’s done on the @NVIDIA blog → https://t.co/8LGVyI71iU
Try Pix2Pix yourself → https://t.co/5QODTeNppR
The Ramanujan Machine: Automatically Generated Conjectures on Fundamental Constants
Like GIMPS or SETI@home, they launch a distributed platform for finding new formulas for fundamental constants. Found new ways to calculate π, 𝑒:
https://t.co/Nyb7eE9XH4 https://t.co/RTWJaLPMmZ
#pico8 💎
e={3,11,5,8,14,2,9,10,4,13,7,6}
::_::
color(1)
for i=0,80 do line(rnd(128),rnd(128)) end
for n=1,10,3 do
a=n/4-t()/4
x=64+cos(a)*42
y=64+sin(a)*42
for j=-1,1,.02 do
i=flr(j+t()*3)
line(x,y+20,x+j*20,y,e[n+i%3])
color(e[n+(i+1)%3])
line(x+j*10,y-10)
end
end
flip()goto _
Wave Physics as an Analog Recurrent Neural Network
They show that wave-based physical systems can be trained to operate as an RNN, and can passively process information in their native domain, without analog-to-digital conversion.
https://t.co/hxNQKIlR24 https://t.co/A910ZCsmKU
Releasing the Sparse Transformer, a network which sets records at predicting what comes next in a sequence — whether text, images, or sound. Improvements to neural 'attention' let it extract patterns from sequences 30x longer than possible previously: https://t.co/FZlDEPsi1A