Excellent work!
I have a quick question: how long did it take you to train it? On my end, training one epoch on a 24GB 4090 takes over ten hours, which is crazy!
By the way, I can only set the batch size to 4. I've scaled down the learning rate proportionally, but will this still significantly affect the training results?
Excellent work!
I have a quick question: how long did it take you to train it? On my end, training one epoch on a 24GB 4090 takes over ten hours, which is crazy!
By the way, I can only set the batch size to 4. I've scaled down the learning rate proportionally, but will this still significantly affect the training results?