Decided to train a few more epochs to see what happens. The result is a slight improvement, mostly in the background, which is now more detailed.
This is epoch 11, the other epochs can be downloaded from tensor.art/models/837001590532518465/Virginia-Sterrett-1-1024CpD6C5-2025-03-05-01:45:56-Ep-8
FLUX.1 - dev-fp8
Trigger: sterrett1 illustration
Repeat 20 Epoch 8 (Trained on 19 1024x1024 images for 3040 total steps)
Then trained for another 3 epoch for total of 11 epoch (4180 total steps)
Unet LR: 0.0005 Scheduler: cosine Optimizer: AdamW
Network Dim: 6 Alpha: 3
Epoch Loss
1 0.378
2 0.382<
3 0.363
4 0.364<
5 0.366<
6 0.356
7 0.354
8 0.347<-lowest
Prompt: sterrett1 illustration of woman on a horse, smoking.
9 0.360
10 0.351<-lowest
11 0.353
Decided to train a few more epochs to see what happens. The result is a slight improvement, mostly in the background, which is now more detailed.
This is epoch 11, the other epochs can be downloaded from tensor.art/models/837001590532518465/Virginia-Sterrett-1-1024CpD6C5-2025-03-05-01:45:56-Ep-8
FLUX.1 - dev-fp8
Trigger: sterrett1 illustration
Repeat 20 Epoch 8 (Trained on 19 1024x1024 images for 3040 total steps)
Then trained for another 3 epoch for total of 11 epoch (4180 total steps)
Unet LR: 0.0005 Scheduler: cosine Optimizer: AdamW
Network Dim: 6 Alpha: 3
Epoch Loss
1 0.378
2 0.382<
3 0.363
4 0.364<
5 0.366<
6 0.356
7 0.354
8 0.347<-lowest
Prompt: sterrett1 illustration of woman on a horse, smoking.
9 0.360
10 0.351<-lowest
11 0.353