ParentFull threadatdrummond·It could be a positive for GPU capability - such as models that would be constrained on GPUs of say 24 GB.This will be a positive development for LLMs like Llama.View on HN