Hey thanks for letting me know.

Its interesting that A1111 uses the name from the metadata instead of the filename. I guess the only solution is to edit the metadata of my Loras whenever I change the filename.

.

If you are interested in the details of why this happens:
This happens whenever I train a lora with a certain name and then rename it for the release. In this case the lora was trained as SerenaPokemonIXL and it made two epochs. I named epoch 10 "SerenaPokemonIXL_v1" and epoch 09 "SerenaPokemonIXL_v2."
Recently I've been doing a lot of experiments with epochs so this unsynchronized metadata name and filename is going to be prevalent on most of my new loras. 😅

@ImStarWars I wouldn't worry about too much about this being perfect because Lora file names can be changed to whatever by the user. If you are making Loras and its something you want to fix I'm pretty sure you can just open the .safetensors file in a text editor and edit the "__metadata__" object. I've never manually edited the metadata but theoretically it should be as easy as editing the text. The data in a safetensor is huge but my guess is you can search for the field "ss_output_name" and just change it.
Anyway like I said, I would suggest focusing on learning other aspects of Lora creation. Understanding how Network Dim/Alpha, # of images, # of repeats per image, and batch size effect your lora are much more useful things to learn. There are a lot of other things to learn but these are the main things that I adjust.
If you ever have questions about Lora training feel free to ask! I'm always happy to provide info/tips if I have time 😁