When you look at the variables in the pretrained base uncased BERT the varibles look like list 1. When you do the training from scratch, 2 additional variables per layer are introduced, with suffixes adam_m and adam_v. It would be nice for someone to explain what these variables are? and what is their significance to the process of training?
If one were to manually initialize variables from a pri

I found that yapf doesn't recognize the SPACES_AROUND_SUBSCRIPT_COLON knob that's listed in the docs. Here's how I produced the error:

Config (setup.cfg)

[yapf]
based_on_style = pep8
spaces_around_subscript_colon = true

Shell

>> yapf -i <filepath>
yapf: Unknown style option "SPACES_AROUND_SUBSCRIPT_COLON"

"scipy.misc.imsave" was removed in scipy version 1.3
--> pip install scipy==1.2.1

https://docs.scipy.org/doc/scipy/reference/release.1.3.0.html#scipy-interpolate-changes

and functions from scipy.misc (bytescale, fromimage, imfilter, imread, imresize, imrotate, imsave, imshow, toimage) have been removed.

I made up a guide to setup Firebase services into a real-world scenario with cutting-edge technologies. I think you could get some insights on DX of your services and how they are used out in the wild (or add it to this examples repository).
Tech used: Quasar (Vue), TypeScript, Jest (unit testing), Firebase Firestore, Cloud Functions, RxFire.

Inside the guide there are some helpers and code ma

Hi team!
I'm trying to run the API generator for a new API with a private discovery document. However, I'm running into a few issues.

• I can't find documentation on how to run the generator. I think it would be really useful to write this down somewhere in a README. I don't use this package enough to remember the commands.
• I'm not able to generate after a clone, npm i, and `npm run gener

I understand that these two python files show two different methods to construct a model. The original n_epoch is 500 which works perfect for both python files. But if I change n_epoch to 20, only tutorial_mnist_mlp_static.py can achieve a high test accuracy (~0.97). The other file tutorial_mnist_mlp_static_2.py only get 0.47.

The models built from these two files looks the same for me (the s

Ref google-map-react/google-map-react#314 (comment) && #524

Example links in DOC.MD are dead.

This happened because they were hosted in webpackbin, who has been deprecated and merged with [codesandbox.io](https://codesand

So, I'm eager to contribute to microG. There's just one problem: I've spent the last 10 hours trying to decipher the codebase, and feel like I'm nonethewiser for it. Could you please please PLEASE add some documentation? At least the architecture? I understand that nobody likes writing docs, and I understand that much of the structure is forced upon you by having to copy GPS, but it's critical i

In the documentation for the Mobileclient.get_promoted_songs() method, the docs say that the method determining the songs returned is unknown. When I run this method my liked/thumbs up auto-playlist is returned. Not sure if that is the case for anyone else.

I think you could rename the project's file to follow the guidelines with lowercase and undescore. Just an improvement :)

`
Some file systems are not case-sensitive, so many projects require filenames to be all lowercase. Using a separating character allows names to still be readable in that form. Using underscores as the separator ensur

Please rewrite the cloud_firestore library in native dart, so we can remove all the big (in terms of download size) Android and iOS libraries. Which slow down build time dramatically.

Having to create a manual abstraction to support https://pub.dev/packages/firebase and https://pub.dev/packages/cloud_firestore is a mess.

All future platforms would be supported out of the box.

Something l

Is your feature request related to a problem? Please describe.
I couldn't find OpenThread's configurations description in https://openthread.io/reference. I think it helpful to have a centered place to introduce all OpenThread configuration and usage.

Describe the solution you'd like
Add a generated configurations page.

Google suggests enabling gzip compression before transferring files over the network:

In order to receive a gzip-encoded response you must do two things: Set an Accept-Encoding header, and modify your user agent to contain the string gzip. Here is an example of properly formed HTTP headers for enabling gzip compression:

Acc

Hello, first of all, a big thank you! For ages I had been looking for a good opensource convenient note app, and since I found yours, I'm just so happy. Great design, simple to use, and powerful also!

A good app is one you use so much that you want to contribute and improve it :) so here are a few questions-suggestions. I'll post them in various issues, so maybe it can be simpler to track.

I