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Outdated suggestions cannot be applied.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.
This PR replaces OpenSora's T5 with the one from
mindone.transformers
.Differences between the refactored and the previous T5:
T5LayerNorm
,hidden_states
are now calculated ashidden_states = hidden_states * ops.rsqrt(variance + self.variance_epsilon)
instead ofhidden_states = hidden_states / ops.sqrt(variance + self.variance_epsilon)
previously.T5DenseActDense
, a custom GELU activation is used (NewGELUActivation
), instead ofnn.GELU()
used previously.T5Attention._relative_position_bucket
,relative_position_if_large
is calculated as follows to align with the PyTorch implementation:If the previous and current T5s are aligned (considering the points mentioned above), the outputs will be identical in both Pynative and Graph modes for all data types, i.e., FP32, FP16, and BF16.
Error compared to HF's
transformers
on 6 random long text samples: