Version Update : v3

This version uses a different base model along with additional datasets. The model has been improved and is now more stable compared to previous versions. However, please avoid using overly complicated or conflicting prompts, as they may make the model difficulty interpreting them correctly.

V3 was trained using the same dataset as BCute Anima, so the two models may feel slightly similar, although they use different architectures.

Thank you for trying the model and supporting the project!