
Update: Versions 1 and 2 are SD 1.5 based models, where as Version XL1.0 is an SDXL version. Please note that there are different trigger phrases for each
LoRA of Bollywood starlet Ananya Panday. Trained on Kohya SS, using images available through search engines. Recommend weight of 0.6-0.7, also CFG between 5.5-7, and 20-40 steps (I've found an inverse correlation between the CFG and steps).
描述:
Learned a lot since I created the V1 LoRA, which was my first ever.
This is trained with the same images as the first version, but with fewer training steps, and with regularization images. The end result is better pictures, for the same prompts.
Recommended settings:
CFG: ~7
Steps: 20-30
Sampler: Euler a
Weight:0.6-0.8
训练词语: ananyapanday
名称: ananyapanday_v2.safetensors
大小 (KB): 147571
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success