This repository was archived by the owner on Feb 13, 2025. It is now read-only.
Implement the DLRM model #344
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
|
Thanks for your pull request. It looks like this may be your first contribution to a Google open source project (if not, look below for help). Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA). 📝 Please visit https://cla.developers.google.com/ to sign. Once you've signed (or fixed any issues), please reply here with What to do if you already signed the CLAIndividual signers
Corporate signers
ℹ️ Googlers: Go here for more info. |
|
@googlebot I signed it! |
|
CLAs look good, thanks! ℹ️ Googlers: Go here for more info. |
dan-zheng
reviewed
Feb 27, 2020
xihui-wu
reviewed
Feb 28, 2020
|
Thanks for contributing this @jordannad. Two general questions regarding the model as I looked into PyTorch implementations:
|
xihui-wu
reviewed
Feb 28, 2020
As part of ensuring the train test is reliable, I have matched the initialization of the reference implementation for the Embedding layers. Additionally, the DLRM model is susceptible to a "bad initialization" that doesn't perfectly memorize the single test minibatch. Although this is infrequent (~1 out of 50 test runs), I have modified the tests to randomly re-initialize 5 times, ensuring the test is approximately flaky with a probability of 3.2e-9 while still maintaining the quality of the test (e.g. testing random initialization, etc). Finally, instead of checking that loss drops below a particular value, the test checks that the accuracy is 100%. This results in a faster stopping condition, and thus the convergence test often runs in under 300ms on a laptop.
|
Please take a look? I believe I've addressed the comments. Thank you! |
xihui-wu
reviewed
Mar 25, 2020
Models/Recommendation/DLRM.swift
Outdated
| sparseInput: [Tensor<Int32>] | ||
| ) -> Tensor<Float> { | ||
| precondition(denseInput.shape.last! == nDense) | ||
| assert(sparseInput.count == latentFactors.count) |
There was a problem hiding this comment.
use precondition as well ?
xihui-wu
approved these changes
Mar 25, 2020
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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.
No description provided.