Open CDchenlin opened 1 month ago
I'd also be interested to know how long the training took on which hardware. The paper says that the model was trained for 250 + 100 epochs with an epoch length of 72000 and that inference for one image took 0.3 seconds on a V100 GPU (x2 for backpropagation). That gives (250 + 100) * 72000 * 0.3 sec * 2 = 175 days
for one V100 GPU. Is this in the right ballpark?
Thank you for your excellent work. I was wondering if there are any plans to release the training or fine-tuning code?