Uploaded by relgukxilef
1200x838 PNG 2.23 MB
Interested in advertising on Derpibooru? Click here for information!
Help fund the $15 daily operational cost of Derpibooru - support us financially!
Description
https://derpibooru.org/869289 run through GoogLeNets deep dream script.
Tags
+-SH safe2175979 +-SH artist:aquagalaxy467 +-SH artist:php13038 +-SH edit173152 +-SH princess luna117241 +-SH alicorn314782 +-SH pony1604204 +-SH g42030832 +-SH cloven hooves16622 +-SH collaboration6962 +-SH curved horn11466 +-SH deepdream64 +-SH female1804484 +-SH galaxy825 +-SH horn191342 +-SH inceptionism11 +-SH planet1632 +-SH solo1427324 +-SH space6563 +-SH stars23435 +-SH surreal1128 +-SH unshorn fetlocks46884
Loading...
Loading...
How’s that going? It sounds really awesome!
I’ve been trying to get the deepdream scripts to run, but was having stupid problems with python finding my multicore caffe build.
Anyway, if you get some cool results with the new data, I can’t wait to see it!
Working for the last couple of days on an ANN with the model of the GoogLeNet trained on Derpibooru images. I’m using 62,000 images as of now. (Sorry Derpibooru!)
Huh… Well I got as far as installing Anaconda.
Maybe there’s a way to figure it out. Because this looks like a sweet thing and I’d at least like to try and act on the challenge to run a scene of MLP through this. I had been selecting .gifs to see if those’ll work at all.
I’m running it in a VM with Ubuntu on a Windows 7 host. You might have trouble getting CUDA to work in the VM though.
I looked at Cafe’s prerequesites and saw a bunch of stuff about Linux and got unsure. I should be able to pull this off on a Windows machine then?
Took my AMD Phenom II X6 1055T processor 5 minutes but if you have CUDA you’ll be golden! Had to downscale the image though because it ran out of memory (8 GB) otherwise.
How intensive is this, if at all? I want to play with it.