nzpeng
Results
2
issues of
nzpeng
大佬的代码简洁至极!发现一些问题:ADM infer部分,当前预测的dt_predict 为小数,而非整数,直接拼在p_code上,作为下一步的输入。这种方式貌似和训练时输入都是整数不相符,有可能出现意想不到的结果。建议改为: dt_predict = torch.round(dt_predict).clamp(1, self.max_duration_token)
Hi, I found the implementation of ResidualCouplingLayer.forward(normalize_flow.py) is different from official VITS code, and this section is not described in the VITS2 paper. what principle your implementation is based on?...