CompressAI
CompressAI copied to clipboard
Question about the usage of Gain in vbr.py
I notice that you have implemented the variable rate model. It's awesome.
While I have some questions about the Gain. I know that the lambda is used to control the rate-distortion in the past. While what is the function of Gain. How does Gain get there?
self.Gain = torch.nn.Parameter( torch.tensor( [0.10000, 0.13944, 0.19293, 0.26874, 0.37268, 0.51801, 0.71957, 1.00000] ), requires_grad=True, )