Add the Adagrad with Momentum optimizer. #269
Merged
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
![copybara-service[bot]](https://avatars.githubusercontent.com/in/44061?s=60&v=4){kind=small}
copybara-service
bot
force-pushed
the
test_765046416
branch
4 times, most recently
from
9d061e1 to
50adf60
Compare
copybara-service
bot
force-pushed
the
test_765046416
branch
from
50adf60 to
1fff875
Compare
copybara-service
bot
changed the title
copybara-service
bot
force-pushed
the
test_765046416
branch
from
1fff875 to
b0703e5
Compare
This implementation matches the one described in [Duchi et al., 2011](https://arxiv.org/abs/1103.4296) with the momentum integration discussed in [Sutskever et al., 2013](https://proceedings.mlr.press/v28/sutskever13.pdf). Features / Params in `AdagradMomentumOptimizerSpec`: - **learning_rate**: The base learning rate. - **momentum**: The momentum parameter (exponential decay for the momentum buffer). - **beta2**: The decay rate for the running average of squared gradients (accumulator). - **epsilon**: A small constant added for numerical stability. - **exponent**: The power to which the accumulator is raised (often referred to as `k_power` in some contexts, typically 0.5 for Adagrad). - **use_nesterov**: Boolean flag; if true, Nesterov momentum is used. The optimizer maintains two slot variables for each trainable embedding parameter: - **accumulator**: Stores the running average of squared gradients. - **momentum_buffer**: Stores the first-order momentum term. PiperOrigin-RevId: 769858905
copybara-service
bot
force-pushed
the
test_765046416
branch
from
b0703e5 to
191e114
Compare
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Outdated suggestions cannot be applied.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.
Add the Adagrad with Momentum optimizer.
This implementation matches the one described in Duchi et al., 2011 with the momentum integration discussed in Sutskever et al., 2013.
Features / Params in
AdagradMomentumOptimizerSpec:k_powerin some contexts, typically 0.5 for Adagrad).The optimizer maintains two slot variables for each trainable embedding parameter: