@aurita yuss I'm on it now! besides the new cool UI update I haven't really used any of the new features, but gpu rendering got improved I think which is cool?!!

@calv oh ym god I am sending u my strength I'm so sorry. I'm just happy ur not using eclipse!!

moved back to sublime text after using vscode for a year and oh my god this is SO much better I missed youuuuuu smooch smooch smooch

@ocean we have to PAY TAX every single day and don't even get scritchies in return what is the point

I hate credit score!!!! I want to be a cat!! AGONY

I've decided that mining sux!! I think itd be way cooler to have "Something from nothing" rather than taking up the earth's resources.. So hear me out?! .. you can decompose things u grow into compost, which you can sift through for rocks or teeny tiny scraps of iron. This can be automated and gives u the incentive for super big optimal farms. By proxy, literally everything is made of carrots lol

Show thread

IM. GLAD PEOPLE ARE RELATING TO THIS BUT ALSO HUGS U, UR ALL DOING GOOD :bunSadPat: :bunSadPat:

Show thread

@pry :blobcatSmileHappy: :blobcatSmileHappy: :blobcatSmileHappy: :blobcatSmileHappy: pry wtf thank you YOURE cool!! i get jealous of the cool stuff you do a lot ahaha thank you! AND GOOD LUCK, A C PROJECT WOULD BE DOPE

@pry Heck yess it's so insanely cool and elegant right!! One of my favourite ideas ever. Like just the idea of like "oh just have a Wrongness function, derive it, and go down the gradient so it becomes less wrong"

@pry OH sorry!! It's a fully connected multi-layer perceptron. One in Rust would be freakin sweet!!

@twinkle ahaha, i love the easy debugging tools!! it's just what im used to after getting an industry job :blobcatMeltMeltMelt:

@pry YA literally! I did that and wrote a custom malloc function that plays with that single pre-allocated block. I've read that a lot of hardcore coders make entire games in just one big malloc too, apparently it makes things a lot easier and faster

@pry The entire network, nodes, synapses etc are all in a single contiguous memory block, which means when that memory gets pulled to CPU cache, its wayyy less likely to have cache misses.

For example if I coded this in C++ and used std::vectors for each layer of nodes, then memory *could* be split across the heap, causing a ton of cache misses! That's a very very basic rundown ahaha

Finished my attempt at a super-cache-coherent C neural network! it's a bit simple but it was fun revisiting it ;u; github.com/tffdev/CNeuralNet

@aurita AAAAAAAAAAAAAAAAAAAOOOOOOOOOOAOOOOOOOOOOOAAAAAAAAAAAAAAAAAAAOOOOOOOOOAAAAAAAA<3333333 :blobcatOwoEyes: :blobcatOwoEyes: :blobcatOwoEyes: :blobcatOwoEyes: :blobcatOwoEyes: :blobcatOwoEyes: :blobcatOwoEyes: :blobcatOwoEyes: :blobcatOwoEyes:

Show older
RaRuRe

It's pronounced ʁaʁyʁe. And written RaRuRe, or R3 for short.
You can find more infos on this place by clicking there.