use biased estimate of std in layernorm as in the original paper #119
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The original paper computes a biased estimate of sample standard deviation. However, by default,
torch.Tensor.std()
uses an unbiased estimate Ref. Therefore, it is necessary to usetorch.Tensor.std(-1,unbiased=False)
. Moreover, the classnn.LayerNorm()
uses biased estimate as well. Though it does not make much difference for large dim, following the definition given in the cited paper is more appropriate.For PyTorch>=2.0, use
torch.Tensor.std(-1,correction=0)
.