Proposal: uncensoring LoRAs

On huggingface there are millions of "models" which are uncensored versions of other models. I find it a huge waste of space. It would be better to post the LoRA of the abliteration/uncensorship. The advantage: a lora can be applied to any model of the same architecture. And to any quant. An example: You can apply THIS to any Qwen3.6-35B-A3B version or quant. I personally think if would be great to have a repository with one lora for each model. Perhaps we could then rank them more easily. What do you think?