Fix mixed precision in TF models #9163
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| @@ -56,14 +57,20 @@ def mish(x): | |||
|
|
|||
| def gelu_fast(x): | |||
| x = tf.convert_to_tensor(x) | |||
| coeff1 = tf.cast(7978845608, x.dtype) | |||
| coeff1 = tf.cast(0.7978845608, x.dtype) | |||
There was a problem hiding this comment.
wow was that wrong the whole time before?
There was a problem hiding this comment.
Yep! I was as surprised as you 😄
patrickvonplaten
approved these changes
Jan 20, 2021
sgugger
approved these changes
Jan 20, 2021
src/transformers/activations_tf.py
Outdated
Comment on lines
66
to
74
| if version.parse(tf.version.VERSION) >= version.parse("2.4"): | ||
| gelu = tf.keras.activations.gelu | ||
| else: | ||
| gelu = tf.keras.layers.Activation(_gelu) |
There was a problem hiding this comment.
Nice to be able to use the TF version now!
LysandreJik
approved these changes
Jan 21, 2021
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
What does this PR do?
This PR aims to fix the mixed precision issues when
tf.keras.mixed_precision.experimental.set_policy()is set to something else thantf.float32. In the same page, this PR aims to fix some TFLite quantization issues.Before to further continue this PR, the PR #9418 has to be merged.
Fixes # (issue)
#7052