> using multiple accounts to work around access or resource usage restrictions

me when training lon ilo nanpa mi

rate my deep learning setup

No shot 😩

instead of training on fp16 this uses nanpa|||

Is it possible to run the tacotron2 synthesis locally without colab? Tried it but wasn't sure on how to modify the colab code to work.

you can try one of these (requires conversion which you might need to learn python for)

Guess I give up on seeing if my latest models are in Uberduck but I can give you the IDs for the models to use in the talknet synthesis notebook. Bob The Tomato (2000s): 19SMFtBNZtuu9C1hOcpsfV8pyS7cm0Zy3 Larry The Cucumber (2000s): 19_KR7Z0bpTOG16jW-_g-_XjVF6-AeXaQ Have fun with it! 😉

Still doing 16+ hour long sessions with DeepFaceLab 0 issues, some have reported that their notebooks are getting flagged as against tos but we will see

Still unclear as to what they are trying to say by "deepfakes" because faceswapping is using way more resources than voice fakes but both most likely fall under the term deepfake

>

fucking stop me google

Pro users seem to not get hit at all rn but still weird situation

also not allows RDP control on Google Colab

and download torrents and share multimedia files on google colab

Literally 1984

and they disallowed entirely from creating deepfakes according of colab updated rules

luckily i dont think tacotron, other TTS, and jukebox don't count as deepfakes currently since nothing is being said when you run them

and if they ban me then i will sue them for money

Interestingly, the change was made between the 14th and the 24th, on the 17th a article was published by VICE bashing innocent deepfake creators and DeepFaceLab contributers because 1 NSFW site was linked as a resource on the github page, even with a nsfw warning

Why would that not make them count as deepfakes

Also I have to say that DFL users train for like hundreds of hours on colab to get a decent model so sorry if we are taking the t4s and p100s

i get t4's all the time so it doesn't effect me lol

runtime exceeding 2 hours detected activating nuclear fission warhead

Personally I hate the t4 because its slow especially when you pay for it lol

i understand why they're doing it but it's still a bummer

knowing google its probably because "deepfakes = gpu usage = no stonks"

I want to know if its resource or ethical stuff

theyve done stuff relating to resources before