ModelsLab/Sagging [Flux] - V0.1.10

Dataset:Only closeups, to short captions, fragmented into tags due to the incorrect use of the onsite trainer from Tensor.art.Known issues:Simple, short prompts work better than complicated, long ones.The subject cannot assume all poses, for example it cannot lie down.The subject does not always know where the front and back are. As a result, the anatomy is sometimes considerably disturbed (bottom on the knees or stomach, etc.).Simple poses in a standing position mostly work well.Stay tuned. v0.2 - v0.9 will be published within the next few hours.Details:Images: 60Captions: Poor, natural language, destroyed from TAEpoch 10 · Steps: 6000 · Loss: 0.312Trigger words: -Repeat: 10Epoch: 10Save Every N Epochs: 1Clip Skip: -Text Encoder learning rate: 0.00001Unet learning rate: 0.0001LR Scheduler: constantOptimizer: AdamW8bitNetwork Dim: 16Network Alpha: 16Gradient Accumulation Steps: -Noise Offset: 0.03Multires noise Discount: 0.1Multires noise iterations: 10conv_dim: -conv_alpha: -Batch Size: -Sampler: euler