LoRa, which I trained on my own computer, is still working on how to adjust parameters and improve the data set.
The current version is 0.1, waiting to be fully perfected.
Description
Some dataset improvements were made and lower learning rates and larger amounts of synthetic data were used, but the results were not good enough to accurately match all the characteristics of the characters.
Fine-tuning models using other illustrious models do not adapt well.
I am now considering training other comic characters and various elements to improve my related experimental data and cognition.