Trained for 7 epochs and 16800 steps.

Trained for 7 epochs and 16600 steps.

Trained for 7 epochs and 16400 steps.

Trained for 1 epochs and 9800 steps.

Trained for 91 epochs and 9600 steps.

Trained for 0 epochs and 9600 steps.

Trained for 0 epochs and 9500 steps.

Trained for 0 epochs and 9000 steps.

Trained for 0 epochs and 8000 steps.

Trained for 0 epochs and 7000 steps.

Trained for 0 epochs and 6000 steps.

Trained for 0 epochs and 5000 steps.

Trained for 0 epochs and 4000 steps.

Trained for 0 epochs and 3000 steps.

Trained for 0 epochs and 2200 steps.

Trained for 0 epochs and 1800 steps.

Trained for 0 epochs and 1600 steps.

Trained for 0 epochs and 1400 steps.

Trained for 0 epochs and 1200 steps.

Trained for 0 epochs and 800 steps.

Trained for 0 epochs and 600 steps.

Trained for 0 epochs and 400 steps.

Trained for 0 epochs and 400 steps.

Trained for 0 epochs and 200 steps.

Trained for 0 epochs and 6200 steps.

Trained for 0 epochs and 6100 steps.

Trained for 0 epochs and 6000 steps.

Trained for 0 epochs and 5900 steps.

Trained for 0 epochs and 5800 steps.

Trained for 0 epochs and 5700 steps.

Trained for 0 epochs and 5600 steps.

Trained for 0 epochs and 5500 steps.

Trained for 0 epochs and 5200 steps.

Trained for 0 epochs and 5100 steps.

Trained for 0 epochs and 3900 steps.

Trained for 0 epochs and 3800 steps.

Trained for 0 epochs and 3700 steps.

Trained for 0 epochs and 3600 steps.

Trained for 0 epochs and 3500 steps.

Trained for 0 epochs and 3400 steps.

60600 + 15400 steps of combined MJ/LAION data = 76000 steps total. used approx 3000 steps of bias on timesteps 400-1000

38800 + 21800 steps MJ 5.2 (a little more than one epoch) = 60600 steps

38800 steps, mj 5.2, no min-snr-gamma, zsnr v-prediction, XL architecture