(Recommended settings below)
I do not own this model. This is a rehost. The original is from Civitai user Defozo (https://civitai.com/models/701671/nsfw-master-flux-lora-merged-with-flux1-dev-fp16), who merged Civitai user Shopon_Skp's NSFW LoRA and Flux Dev fp16 checkpoint. If you are any of these people and do not wish to have this uploaded on TensorArt, please contact me. Thank you
Original description:
This is a merged model of a base model flux1-dev fp16 and NSFW MASTER FLUX LoRA: https://civitai.com/models/667086?modelVersionId=746602To merge, I used a script from:https://github.com/kohya-ss/sd-scripts/Using --ratios 0.8 in the merging command.
Steps: 20,000
Epochs: 20
RECOMMENDED SETTINGS:
If you want to use a method that most use with advanced options on, I recommend:
sampler as euler,
scheduler as simple
Another one that others use is:
sampler as dpmpp_3m_sde_gpu,
scheduler as karras
For sampling steps, I recommend 30 for extra LoRAs, 20 for base model. Guidance scale at 4 is good too.