Nice!
Fortunately MTEB lets you sort by model parameter size because using 7B parameter LLMs for embeddings is just... Yuck.