PURE is the first framework to employ a multimodal large language model (MLLM) as its backbone for addressing low-level vision tasks.

We provide the implementation of PURE, as well as sampling code, in our github repository.