LoRa & LyCORIS Fusion
Trained on ~217 650 MJ Images
LyCoris trained on 211 924 MJ Images
LoRa trained on 217 650 MJ Images
Merged in di.FFUSION.ai-stage-I.safetensorsdim
dim128 | conv32
UNet attention weight average magnitude: 5.714574005824907
UNet attention weight average strength: 0.014448476696414102
UNet conv weight average magnitude: 3.623326885917904
UNet conv weight average strength: 0.00653153534681701
Text Encoder weight average magnitude: 5.401999128837191
Text Encoder weight average strength: 0.010774937006810953
Training took less than a day on both using a 8 a100 (80GB) and 8 40GB
Samples were generated on SD 2.1 768 NoEma & Ema / realismEngine_v10 / classicNegativeSD21_classicNegative768px / aZovyaRPGArtistTools_sd21768V1
Negative used:
(deformed, distorted, disfigured:1.3), poorly drawn, bad anatomy, red eyes, wrong anatomy, extra limb, missing limb, floating limbs, (mutated hands and fingers:1.4), disconnected limbs, mutation, mutated, ugly, disgusting, blurry, amputation (oversaturated:1.5),(saturated:1.2),big contrast,contrast white burn,white spots overexposed,over saturated,burnt image,ugly,hideous,deformed,mutant,butchered,gore,sloppy,artifacts,mutilated,poorly drawn,poorly detailed,smudged,sketch,pencil,glossy skin,doll,plastic,lowres,overexposed,broken,fuzzy,aliasing,cheap,oldschool,poor quality,pixelated,sleepy,closed-eyes,crippled,crooked,weird,odd,distorted,erased,cut,hideous,pixelated,text,logo,watermark,deformed eyes,close up,((disfigured)),((bad art)),((deformed)),((extra limbs)),(((duplicate))),((morbid)),((mutilated)),out of frame,extra fingers,mutated hands,poorly drawn eyes,((poorly drawn hands)),((poorly drawn face)),(((mutation))),blurry,((bad anatomy)),(((bad proportions))),cloned face,body out of frame,bad anatomy,gross proportions,(malformed limbs),((missing arms)),((missing legs)),(((extra arms))),(((extra legs))),(fused fingers),(too many fingers),(((long neck))),tiling,mutated,cross-eye
Training data:
FFusion SM - LyCoris
"ss_sd_model_name": "stabilityai/stable-diffusion-2-1-base",
"ss_resolution": "(768, 768)",
"ss_clip_skip": "2",
"ss_num_train_images": "211924",
"ss_network_args":
"conv_dim": "16",
"conv_alpha": "16",
"algo": "loha"
"ss_network_module": "lycoris.kohya",
"ss_optimizer": "torch.optim.adamw.AdamW(weight_decay=0.01,betas=(0.9, 0.99))",
FFusion SMB 3 - LoRa
"ss_resolution": "(512, 512)",
"ss_clip_skip": "2",
"ss_num_train_images": "217650",
"ss_network_alpha": "128.0",
"ss_network_args": {
"conv_dim": "32",
"conv_alpha": "32"
"ss_network_dim": "128",
"ss_network_module": "networks.lora",
"ss_num_batches_per_epoch": "54422",
"ss_optimizer": "dadaptation.dadapt_adam.DAdaptAdam(decouple=True,weight_decay=4.0,betas=(0.9, 0.98),use_bias_correction=True,log_every=100)"
Fully trained only with Kohya S & Shih-Ying Yeh (Kohaku-BlueLeaf) tools 🤟 🥃
Description
FAQ
Comments (4)
Will this work with a 1.5 checkpoint?
Unfortunately, it's designed exclusively for 2.x versions.
... However, there is a possibility of partially converting and making slight modifications to the text encoder from the model.
Although, it's worth mentioning that there's already an abundance of top-notch 1.5 quality LoRa models available, whereas options for SD 2 are relatively scarce. So, let's embrace the exciting advancements of the 2.x era! 🤟 🥃
@idle Oh yeah, I'm all about the bleeding edge...
I have a few 2.1 models, but I also have 115 1.5 ones. Just a shame, as the LoRa looks amazing 😍
@SerenityX I have to admit that it can be a little ¯\_(ツ)_/¯ unstable due to the large number of images... really pushing it here with those numbers! will continue to share more DyLo and Lyco 2.1 LoRa models for further testing later on
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.