-
Notifications
You must be signed in to change notification settings - Fork 607
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
float16 support for GPU als model (#661)
This adds support for using float16 factors in the GPU version of the ALS model. This reduces the memory needed for the ALS model embeddings by half - while providing a small increase to training time, and virtually no difference in the accuracy of the learned model. All computations are still performed using float32 - including both training and inference. This is done with using mixed precision matrix multiplications during inference : the fp16 factors are multiplied together with results accumulated as fp32. During training, the factors are converted from fp16 to fp32 - and updates are calculated in 32-bit before being stored back as fp16.
- Loading branch information
Showing
17 changed files
with
401 additions
and
145 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
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
Oops, something went wrong.