1) According to step 3, we need to run the hyper-parameter script to find the beta and N. However, it is not clear what is N: in the paper, it used for the amount of samples in the training process, but it seems that for the hyper-paremeter tunning it corresponds to another thing. Is N denoting the inference steps?
2) When I run the inference code, I have a mismatch with the SNR: it seems that it is shifted 15dB wrt the results in the paper. Is there something to change?
Hi, I have the following two observations:
1) According to step 3, we need to run the hyper-parameter script to find the beta and N. However, it is not clear what is N: in the paper, it used for the amount of samples in the training process, but it seems that for the hyper-paremeter tunning it corresponds to another thing. Is N denoting the inference steps?
2) When I run the inference code, I have a mismatch with the SNR: it seems that it is shifted 15dB wrt the results in the paper. Is there something to change?
Thanks in advance!