First, this is my first published lora with larger neural network, I’ve made a lot of tests runs to get better details on complex concepts, releasing this simple one trying to see how it works out in real generation.
Aside from that, the dataset for this lora has been prepared like my typical lora. It is captioned in natural language without trigger words. It is mostly trained on close-ups with some wider angle shots in there to prevent view biases. All faces in dataset have been censored in various ways to prevent the result from looking like the persons in the dataset. If you want to see a face, put details of it in your prompt. Random edits were used to prevent the introduction of bias caused by these edits.
Description
{
"engine": "kohya",
"unetLR": 0.0005,
"clipSkip": 1,
"loraType": "lora",
"keepTokens": 0,
"networkDim": 8,
"numRepeats": 5,
"resolution": 1024,
"lrScheduler": "cosine_with_restarts",
"minSnrGamma": 5,
"noiseOffset": 0.1,
"targetSteps": 2800,
"enableBucket": true,
"networkAlpha": 64,
"optimizerType": "AdamW8Bit",
"textEncoderLR": 0,
"maxTrainEpochs": 40,
"shuffleCaption": false,
"trainBatchSize": 3,
"flipAugmentation": false,
"lrSchedulerNumCycles": 3
}FAQ
Comments (3)
but you dont trained
puffy nipples
its never in your captions, i see it in metadata
The "puffy nipples" trigger word itself isn’t used, tho images of puffy nipples were used, I just used automatic captions which described them as erect nipples. Unfortunately, I do feel like the result isn’t puffy enough or I can’t find a way to get the result I was looking for, I should have been more careful with the captioning. I released it mostly for testing purpose of how a large network dimension lora work in real word.
all fine, maybe more images with head
or try "upper budy" ... (without a lora mostly is without head)
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.