3 ms·
Could this approach be used on other types of models, such as image models eg stable diffusion?
by Seattle3503 4y ago
Could this approach be used on other types of models, such as image models eg stable diffusion?
- ShamelessC 4y agoThere is probably less motivation, as these models are much smaller. An M1 Mac can run inference in under a minute. An GPU in as little as 3-5 seconds. This is supposed to be as much as 20x faster when the distilled Stable diffusion model is released.