The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/3382ba03f380d3f681a84c8c8bf83bebf9ca2b7c

commit 3382ba03f380d3f681a84c8c8bf83bebf9ca2b7c
Author: abakalov <abakalov@chromium.org>
Date: Wed Dec 07 22:15:20 2016

Updating the CLD3 commit hash in DEPS

The main changes are:

1) a retrained model with additional script-based features.
GitHub page showing the diff:
https://github.com/google/cld3/commit/fa5974a4d3b5e7934fcb166ff26ed6bfce68b18a

2) updating the feature spec parser to parse features without args.
GitHub page showing the diff:
https://github.com/google/cld3/commit/c03368eff92acc56756df2d4cd74a01a674409ea

BUG= 672185 

Review-Url: https://codereview.chromium.org/2560473004
Cr-Commit-Position: refs/heads/master@{#437081}

[modify] https://crrev.com/3382ba03f380d3f681a84c8c8bf83bebf9ca2b7c/DEPS

[Automated comment] DEPS changes referenced in bugdroid comments, needs manual review.

Hey Dimu, could you review this if you have a moment. Thanks!

Could you please confirm whether this change is verified in Canary and safe to merge?

Requesting Richard to approve the merge if all looks good.

Hi Ligimole,

This piece of code has been in Canary for 5 days. Can you please clarify what you mean by "verified in Canary and safe to merge" (e.g., are you looking for the output of some specific tools)?

Thanks,
Anton 

We use canary to verify changes haven't caused any breakages/crashes, and in the case of a bug fix verify that it's been fixed, before merging.

In the case of CLD3 if you have data to show better performance, and that there isn't a performance regression.  That would be great too.  But mostly we're just concerned with making sure this won't break anything.

Thanks for the clarification!

As far as I know, there haven't been any crashes/problems since the code was submitted 5 days ago.

This update fixes the following bug about a misprediction:
https://bugs.chromium.org/p/chromium/issues/detail?id=649277

There are a few other open ones assigned to me which were fixed by the previous version of CLD3. This update still makes the correct predictions for those cases. I was planning to close all the bugs once the code is merged into Beta.

We ran experiments on a dataset composed of sentences and a dataset composed of web pages. In both cases, this update achieves a strictly better performance.

Please let me know if you need any additional information or more detailed descriptions.

Hi Richard,

Please let us know if any additional information is needed to approve the merging of this CL.

Thanks in advance!

Anton

SGTM, thanks for confirming!

Merging.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/0263b027555a01ea660cf77eaffdac9981d923bb

commit 0263b027555a01ea660cf77eaffdac9981d923bb
Author: Rouslan Solomakhin <rouslan@chromium.org>
Date: Wed Dec 14 21:57:09 2016

[Merge M-56] Updating the CLD3 commit hash in DEPS

The main changes are:

1) a retrained model with additional script-based features.
GitHub page showing the diff:
https://github.com/google/cld3/commit/fa5974a4d3b5e7934fcb166ff26ed6bfce68b18a

2) updating the feature spec parser to parse features without args.
GitHub page showing the diff:
https://github.com/google/cld3/commit/c03368eff92acc56756df2d4cd74a01a674409ea

BUG= 672185 

Review-Url: https://codereview.chromium.org/2560473004
Cr-Commit-Position: refs/heads/master@{#437081}
(cherry picked from commit 3382ba03f380d3f681a84c8c8bf83bebf9ca2b7c)

Review-Url: https://codereview.chromium.org/2571413002 .
Cr-Commit-Position: refs/branch-heads/2924@{#497}
Cr-Branched-From: 3a87aecc31cd1ffe751dd72c04e5a96a1fc8108a-refs/heads/master@{#433059}

[modify] https://crrev.com/0263b027555a01ea660cf77eaffdac9981d923bb/DEPS

Marking this as fixed given the merge back in December.