DiffusionGemma hits 1,000 tokens per second by ditching word-by-word generation entirely. It just doesn't run on most ...
Don't be deadset on using LM Studio when there is a great alternative, llama.cpp, that does everything you need without the ...