-
Notifications
You must be signed in to change notification settings - Fork 46
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
documentation changes and additions for new features
- Loading branch information
Showing
3 changed files
with
29 additions
and
13 deletions.
There are no files selected for viewing
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
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
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,19 @@ | ||
The performance of the training process is sensitive to different constants called hyperparameters. DPP supports automatic hyperparameter optimization via athe grid search method in order to evaluate performance under many different combinations of hyperparameter values. | ||
|
||
## Training with Hyperparameter Search | ||
|
||
In order to train with hyperparameter search, you must replace the line | ||
|
||
``` | ||
model.begin_training() | ||
``` | ||
|
||
with a modified version containing the ranges of hyperparameters you would like to evaluate: | ||
|
||
``` | ||
model.begin_training_with_hyperparameter_search(l2_reg_limits=[0.001, 0.005], lr_limits=[0.0001, 0.001], num_steps=4) | ||
``` | ||
|
||
Here, you can see that we are searching over values for two hyperparameters: the L2 regularization coefficient (`l2_reg_limits`) and the learning rate (`lr_limits`). If you don't want to search over a particular hyperparameter, just set its limits to `None` and make sure you set it manually in your model (for example, with `set_regularization_coefficient()`). The values in brackets indicate the lowest and highest values to try, respectively. The area in between the low and high values is divided into equal parts depending on the number of steps chosen. | ||
|
||
The parameter `num_steps=4` means that the system will search over 4 values for each of the two hyperparameters, meaning that in total 12 runs will be executed. Please note that larger values for `num_steps` will increase the amount of runs exponentially, which will increase the run time dramatically. |
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