Understand interactive "calculator" for Memory-efficient training on https://training-transformers-together.github.io/

[whaowhao]

Thanks for the great work! On https://training-transformers-together.github.io/ there is this interactive "calculator" for Memory-efficient training, but I have some question about the result

Does this mean my personal desktop [GTX 3090 (24GB vram) and 16Gb ram] can be used to finetune GPT3-175B model? If so that'd be sick, and is there a tutorial on it?