My training takes 40 hours, what should I change to optimize it? enable full bf16 training. enable f

My training takes 40 hours, what should I change to optimize it? enable full bf16 training.
enable fp8 training.
running training /
num train images * repeats / ×: 61
num reg images / : 0
num batches per epoch / 1epoch: 61
num epochs / epoch: 200
batch size per device / : 1
gradient accumulation steps / = 1
total optimization steps / : 12200
Was this page helpful?