Replies: 1 comment
-
|
@babysor 请问作者还关注这个项目么?能否解答一下 |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
我使用一个自己的样本库做为素材,对pretrained-11-7-21_75k.pt进行继续训练。
观察到日志中的Epoch: x/y,似乎是一个进度条。于是决定观察一下进度条走到尽头会发生什么,于是得到了下图。似乎是又开始了一组新的训练。
新训练的目标轮数是160k,似乎是来自22858 * 7 = 160k
原训练进度12143 * 7 = 85k,加上起始的75k,结果也是160k。似乎是训练到160k步时会完成一个什么阶段之类的,于是决定去看一下plots。
结果发现160k那一步完成时的注意力模型,斜线近乎完美
程序每25k会自动保存一个模型,我只获得了100k、125k、150k三个模型,160k步时的那个模型,因为训练没停还在继续训练,现在已经没有了。要找回的话,只能拿150k那个模型,重新训练10k步
因此想请问下题目中的两个问题:
日志中的Epoch: x/y 是什么意思?
是否支持在命令行中指定一个训练步数,训练完成后停止训练?
另外,我在执行预处理时有如下提示:
Some weights of EmotionExtractorModel were not initialized from the model checkpoint at audeering/wav2vec2-large-robust-12-ft-emotion-msp-dim and are newly initialized: ['wav2vec2.encoder.pos_conv_embed.conv.parametrizations.weight.original1', 'wav2vec2.encoder.pos_conv_embed.conv.parametrizations.weight.original0']
You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.
似乎是缺少什么语气感情模型。请问这个模型要从哪里获得?
Beta Was this translation helpful? Give feedback.
All reactions