Fine-tuning with LoRA: create your own avatars and styles(kix.in)
kix.in
Fine-tuning with LoRA: create your own avatars and styles
https://www.kix.in/2023/04/07/sd-lora-finetuning/
2 comments
In my opinion the problem of LoRA vs LoRa is that while the two are in different domains (machine learning vs. wireless transmission), it is still very easy to confuse the two because these two domains are adjacent. For example you could imagine that someone made a chat system that works over LoRa, where you can create your own avatars - which is what I assumed this article was about before clicking the link. Technically the two names are different because of the uppercase A in LoRA, but that is easy to miss.
Microsoft also reused LoRA in the LLM context recently
https://github.com/microsoft/LoRA
sigh
https://github.com/microsoft/LoRA
sigh
The Microsoft LoRA is actually the same concept as the one applied here, the original method was applied to LLMs and then brough to diffusion models recently.
In military uniform, steampunk style, moar fractals