Writing an entry point class which essentially is just -
A diffusion module for generating images from given prompts and sampling parameters.
This class will includes a tokenizer, a diffusion model (possibly distributed across multiple GPUs) (not sure about that), and GPU memory space allocated for intermediate states (aka KV cache). Given a batch of prompts and sampling parameters, this class generates images from the model, using an intelligent batching mechanism and efficient memory management.
At least this is the goal, kinda transferring everything from vllm to diffusion
Writing an entry point class which essentially is just -
A diffusion module for generating images from given prompts and sampling parameters.
At least this is the goal, kinda transferring everything from vllm to diffusion