emikulic·3 anni fa·discussThree epochs means it sees each token three times. The dataset is ~1T like you said.
emikulic·3 anni fa·discussI want flying cars and moving sidewalks like I was promised by The Jetsons. :(
emikulic·3 anni fa·discussTry https://huggingface.co/ehartford/WizardLM-7B-Uncensored and related models. They're not even trained on smut, just the neutered responses were removed before the RLHF stage (IIUC)
emikulic·3 anni fa·discussEmad said SD cost $600,000 to train. I wonder if Midjourney also had to pay that to train from scratch.
emikulic·3 anni fa·discussWhat benefits does the Huggingface diffusers(?) implementation have over A1111?
emikulic·3 anni fa·discussI like to run A1111 in --api mode and write my own script to drive it over HTTP.
emikulic·3 anni fa·discussLast time I looked into this, the answer was "because huggingface transformers and torch.load are written to do it this way"You could absolutely do something streaming, or mmap the weights instead of loading them into system RAM. Just the default interfaces don't.
emikulic·3 anni fa·discussThis used to be the case but nowadays you can just pip install e.g. https://libraries.io/pypi/nvidia-cuda-runtime-cu11
emikulic·4 anni fa·discussI hope you will run your own locally instead of letting Midjourney block and/or mess with your prompts.
emikulic·4 anni fa·discussIf edge devices includes my gamer PC then yes. For apps I'd recommend https://github.com/AUTOMATIC1111/stable-diffusion-webui
emikulic·4 anni fa·discussYou have to create an account and agree to some stuff and be logged in but you're correct that you don't have to pay.
emikulic·4 anni fa·discussSame. I've found the biggest problem was SNAT rewriting the (source) port number. netfilter, by default, doesn't do this. pf does but you can configure it not to.