请问训练到25k的时候注意力线还是没有出来，并且文件多出了一个出来这个是正常的吗？

请问训练到25k的时候注意力线还是没有出来，并且文件多出了一个出来这个是正常的吗？





这样的情况是正确的还是我操作错了

cpu是i59400F，显卡是1505ti，配置GPU是的数值只能配置到6，GPU的占用率已经在80-100%跳动了

Tacotron Training
tts_schedule = [(2, 1e-3, 10_000, 6), # Progressive training schedule
(2, 5e-4, 15_000, 6), # (r, lr, step, batch_size)
(2, 2e-4, 20_000, 6), # (r, lr, step, batch_size)
(2, 1e-4, 30_000, 6), #
(2, 5e-5, 40_000, 6), #
(2, 1e-5, 60_000, 6), #
(2, 5e-6, 160_000, 6), # r = reduction factor (# of mel frames
(2, 3e-6, 320_000, 6), # synthesized for each decoder iteration)
(2, 1e-6, 640_000, 6)], # lr = learning rate

还请大佬给指点指点，谢谢！

Start with the Tacotron Training tts_schedule shown in the issue and compare the attached training outputs, including the attention images and extra file. Trace how training artifacts and attention samples are produced, then determine whether the missing attention line and additional file match expected behavior for the reported step and batch size.

Written by the indexing model from the issue text.