Unet Input question?

[chaowentao]

Thank you for your excellent code. I have a little question about the input of the UNet.

According to the paper, the input is (x, y_noise, gamma_t). But in code, I find that

x_recon = self.denoise_fn(
      torch.cat([x_in["SR"], x_noisy], dim=1), continuous_sqrt_alpha_cumprod
 )

According to the equation, continuous_sqrt_alpha_cumprod is sqrt_gamma. Why use sqrt_gamma rather than "gamma to stand for timesteps? Is there any difference here?

[SocraLee]

Check the paper page3 upleft. They are the same thing.