Closed Ther-nullptr closed 2 years ago
@alexeib
it wont have a much of an effect, but you have to match the feature extractor to the normalization setting
normalize in dataloader -> layer norm in feature extractor no normalization in dataloader -> group norm in first block of feature extractor + feature_grad_mult = 0.1 (rescale feature extractor grads by 0.1)
❓ Questions and Help
Before asking:
What is your question?
In wav2vec2.0 and hubert, the config
task.normalize
is set toFalse
(which means not to normalize the input audio), but data2vec is set toTrue
, and the original paper also mentioned it. Will it have a big effect on experiment result?Code
What have you tried?
What's your environment?
pip
, source):