Add new regression loss function type to FBLearner #21080
Closed
Conversation
This file contains hidden or 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
pytorchbot
added
caffe2
module: pybind
richardxiong
force-pushed
the
export-D15524377
branch
from
2468c62 to
7bd5f8a
Compare
richardxiong
force-pushed
the
export-D15524377
branch
from
7bd5f8a to
c68a58f
Compare
richardxiong
force-pushed
the
export-D15524377
branch
from
c68a58f to
489fb6a
Compare
richardxiong
force-pushed
the
export-D15524377
branch
from
489fb6a to
5f59f7a
Compare
richardxiong
force-pushed
the
export-D15524377
branch
from
5f59f7a to
6f748a7
Compare
Summary: Pull Request resolved: pytorch#21080 Add Huber loss as a new option for regression training (refer to TensorFlow implementation: https://fburl.com/9va71wwo) # huber loss def huber(true, pred, delta): error = abs(true-pred) loss = 0.5 * min(error, delta)^2 + delta * max(error - delta, 0) return mean(loss) As a combination of MSE loss (`x < delta`) and MAE loss (`x >= delta`), the advantage of Huber loss is to reduce the training dependence on outlier. One thing worth to note is that Huber loss is not 2nd differential at `x = delta`. To further address this problem, one could consider adopt the loss of `LOG(cosh(x))`. Differential Revision: D15524377 fbshipit-source-id: 4a7b0f69b859de3ca903f3bb070c318e0d1f298a
richardxiong
force-pushed
the
export-D15524377
branch
from
6f748a7 to
b0eca69
Compare
|
This pull request has been merged in 76a250d. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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Outdated suggestions cannot be applied.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.
Summary:
Add Huber loss as a new option for regression training
huber loss
def huber(true, pred, delta):
loss = np.where(np.abs(true-pred) < delta , 0.5*((true-pred)2), deltanp.abs(true - pred) - 0.5(delta2))
return np.sum(loss)
As a combination of MSE loss (
x < delta) and MAE loss (x >= delta), the advantage of Huber loss is to reduce the training dependence on outlier.One thing worth to note is that Huber loss is not 2nd differential at
x = delta. To further address this problem, one could consider adopt the loss ofLOG(cosh(x)).Differential Revision: D15524377