v1.0
Twin sister of my already existing LoRA, trained on same dataset as latest version. Training params is as follows:
"network_dim": 20,
"network_alpha": 20, #treat dim and alpha as one param. Their individual sizes influence amount of details able to fit inside of lora and their ratio alpha/dim is specifing where your lora will be most active during generation steps (alpha/dim<1 more details, alpha/dim>1 more color blobs on earlier stages)
"learning_rate": 0.0001,
"max_train_epochs": 10,
"resolution": 768,
"repeats": 10,
"caption_dropout": 0.1,
"gpu_index": "0", #you won't be able to use gpu without correct Torch+CUDA pair
"optimizer_type": "AdamW8bit",
"lr_scheduler": "cosine_with_restarts",
"lr_scheduler_num_cycles": 1,
"lr_warmup_steps": 100,
"train_batch_size": 1,
"gradient_accumulation_steps": 1,
"max_grad_norm": 1.0,
"save_every_n_epochs": 1,
"save_last_n_epochs": 4,
"mixed_precision": "bf16",
"gradient_checkpointing": true,
"seed": 42,
"noise_offset": 0.03,
"multires_noise_discount": 0.3,
"timestep_sampling": "sigmoid",
"discrete_flow_shift": 1.0,
Base
Apparently, newly emmerged Anima model is a completly different architecture from sd or sdxl, making it mostly incompatible with many good loras. However, training loras for it is not as hard, as one may think. Standalone trainer app from citron is able to make aroud 5000 steps in 3 hours and is using around 5gb of VRAM. It is not as smooth in installation process as one may wish for, but still managable. Especially if you know/can bother to learn how to correctly install PyTorch for your specific version of CUDA driver of GPU. I myself figured it out just from couple of searches in google, so nothing it is not a rocket science.
Description
FAQ
Comments (10)
why did you train on preview 3 instead of base v1.0?
To be honest it was more for a test. I just didn't pay attention to what model was selected in trainer. Next versions probably will be on newer version.
传说中的“掏大粪”
We need to go deeper!
Just show me, how much! (No, really. If you give examples, anything is possible)
@WannaBeWoman Wish I could xD
@Lopodelas Well, it depends. For me, even a link or mention of a tag can be enough. I know that some extreams of this kink do exist (something like body-glove, when size difference is also involved), but I just don't know, wich one of them whould be interresting to people and what to search for.
@WannaBeWoman Well I supopose the only next level I'd seek would (potentially) be deeper and with grabbing/pinching the pussy/anus-walls from the inside (depending on where the fisting happens) and of course that would need xray/cross-section. I've only found like 1 or 2 images total with that but that was a long time ago
@Lopodelas there are indeed some pictures like that in dataset. I havent tagged them specifically, but automatic tags should include something like "cross-section" or "x-ray". But in this version they probably won't be good, but you definetly can try. Look into dataset itself for ideas.
@WannaBeWoman Oh damn, I was struggling to find those kinds of pictures, care to share?