convoluted connor @notwa@witches.town

this dropout-like thing i'm experimenting with keeps loss and accuracy values pretty close across the training and validation sets. unfortunately, overall performance on the validation set is slightly degraded.
https://witches.town/media/ICB8OCP0smGbYbKcmmI

show gan eye, neigh…

somehow, having a song stuck in my head and trivially being able to bring it up and play it never gets old.

today i woke up with this song, specifically the riff around 3:12
https://www.youtube.com/watch?v=19zFHmz1F6k

err when i say original i mean not the same 100 reposts you might see on birdsite

this is the account: @cats go follow for original cat content 👍

movie name, except technology

the account that just boosts pictures of cats across the fediverse is very healing

tired: how tired was i when i wrote this code?
VERY tired: how tired was i when i decided to delete the two lines that made this code work?

COLD GETTIN' DUMB

YEA

this is one of those things that would be trivial in C, but the rest of the program really wouldn't be. I don't feel up to fighting with C++ templates either. maybe Julia is worth looking into?

ugh I had an idea for a neural network layer but I can't think of a way to implement it simply in numpy, nevermind any tensor based libraries. I wish numpy's shuffle function could shuffle rows independently. maybe my best bet is rolling a hashing function or something?

I should get pico-8 and try making something
*1 year later*
I should get pico-8 and tr

actually I'm not even gonna say anything else I jinx myself

gonna unfollow the fame witch bot at least for now

here's a song by them that exemplifies their eccentricity: https://www.youtube.com/watch?v=IaFA7EUYnxI

the song is so good though. the part at 2:00 in the full version is polyrhythm hell. i miss narasaki and coaltar of the deepers

also i'm still sick :DDDDDDD

realizing i'd have a very interesting set of CWs if i decided to link the first goku sayonara zetsubou sensei OP