convoluted connor @notwa@witches.town

i don't want to think about it though, it's just stressing me out. i think i'm gonna see if there's any games or youtube series to get my mind recalibrated

now that i think about it, this linear optimization problem might be failing because the inputs are just too danged correlated

update: added battletoads to the mix, and stats split for opcodes and their addressing modes.

update: tried an impulse, nearly identical results. suspicious… maybe the plugin is doing something *after* the speaker convolution that gets disabled with the preamp or poweramp?

I thought linear optimization was supposed to be easy!

yeah i can't find the post i'm looking for. oh well, here's the reply anyway:

well i tried learning the matrix being dotted but it seems like there's something fundamentally wrong with my idea. (redundancy? autograd bug?)

not really in the mood to debug this, especially since i realized my initial idea wouldn't work after i had started coding it.

i totally messed up trying to say this lol

looking forward to search by own posts in the next version…

<GlitterBerri> 16,000 more lines of text on the wall~ 16,000 more lines of text...
<notwa> take one down, Oh! My God!! RCP is Hung Up!

huh I completely forgot about this thing I was doing with radial basis functions a couple months ago. I wonder if it could help with one of my current experiments

side note: i have no idea how to actually write perl

< textdump.bin | iconv -c -f euc-jp -t utf-8 | tr '\0' '\n' | perl -Mopen=locale -Mutf8 -MUnicode::Normalize -pe'$_=NFKD($_)=~y/ァ-ヶ/ぁ-ゖ/r'
# sorry not sorry

i wanna take a pic but it's still dark out ;-; i mean besides the ambient snow-reflected light, which isn't quite enough to take a good picture

lol there's like a foot of snow on the table outside... probably pretty powdery though

it's a shame the gameboy wasn't 6502-based, if not just so there could be one gen 1 pokemon for each valid instruction.