TL;DR 388 | The Google Developer News Show
0:00 - Intro
0:19 - Popover API lands in Baseline → https://goo.gle/4a3qs2K
0:44 - Achieving privacy compliance with your CI/CD → https://goo.gle/4a6n0Vi
1:24 - Safely repairing broken builds with ML
1:54 - Improving Gboard language models via private federated analytics
2:37 - Gemini Cookbook → https://goo.gle/44jp9vq
Here to bring you the latest developer news from across Google is Rody Davis.Tune in every week for a new episode, and let us know what you think of the latest announcements in the comments below.
Follow Google for Developers on Instagram → https://goo.gle/googledevs
Watch more The Developer Show → https://goo.gle/GDevShow Subscribe to Google for Developers → https://goo.gle/developers
#Google #Developers
Speaker: Rody Davis; Products Mentioned: Gemini, Web Stories - General;
0:00 · [MUSIC PLAYING] RODY DAVIS: Hi, my name is Rody, and I’m a developer relations engineer on the developer solutions team.
0:08 · This is your weekly update on the coolest developer news from Google.
0:11 · Make sure to check the description for any links or topics mentioned.
0:15 · Now let’s get started.
0:17 · Starting with the web, the Popover API lands in baseline and in all major browsers, which means it’s safe to use in your applications and websites.
0:25 · If you’re unfamiliar with the API, it allows you to promote content to the top layer, add light-dismiss, default focus management with tabbing, and finally, accessible keyboard and component bindings.
0:36 · The Popover API makes it possible to create complete interactions that previously required JavaScript.
0:41 · You can learn more on the web.dev blog.
0:44 · Next, we have two stories to help you with your continuous integration and delivery pipelines.
0:49 · First, on the Google Developers blog, you can read about how to achieve privacy compliance using checks.
0:54 · This integration into CI/CD allows you to move towards automating compliance and privacy standards as part of the release cycle.
1:02 · There are multiple benefits when adopting checks into your application release pipeline.
1:06 · This includes real-time intelligent alerting, understanding data sharing for third-party SDKs, making sure you follow company policies, keeping your Google Play Data Safety section up-to-date, and deploying quickly with confidence.
1:20 · You can learn more on the developers.google.com blog.
1:24 · From Google Research, you can now learn how machine learning can be used to safely repair broken builds as part of a CI/CD pipeline.
1:31 · The DIDACT ML model is used to predict build fixes and repair them automatically.
1:37 · This model, which ran as an experiment for 11 weeks, resulted in 2% reduction in active coding time and developer time after a change was submitted to review as well as a 2% increase in total change throughput throughout the week.
1:51 · You can learn more on the research.google blog.
1:54 · Next, heading over to AI, we can learn about how Gboard has been improving its language models via private federated analytics.
2:01 · Gboard uses language models to improve typing through features like next word prediction, autocorrection, Smart Compose, slide to type, and proofread.
2:10 · To boost Google’s keyboard performance while keeping the user data private, we have worked with language experts to refine its dictionaries, developed novel privacy-preserving techniques based on federated analytics and differential privacy to discover out of vocab words directly on the user’s devices, and employed secure hardware infrastructure for confidential and externally-verifiable server-side data processing.
2:32 · You can learn more about the approaches taken on the research.google blog.
2:36 · And finally, you can learn about some of the new exciting additions to the Google Gemini Cookbook.
2:41 · This GitHub repo contains lots of examples that can be run on Colab, targeting Python developers, or anyone wanting to work with Gemini via REST.
2:49 · Some of the exciting quickstarts include content for the following— system instructions, function calling, embeddings with RAG or Retrieval Augmented Generation, uploading files like audio and PDFs through the Files API, finetuning, JSON mode, streaming, and safety.
3:04 · There are so many more examples not listed, but make sure to check out the GitHub repo.
3:07 · You can open any of them in Google’s Colab and grab an API key from Google AI Studio to get started.
3:13 · You can learn more at github.com/google gemini/cookbook.
3:18 · To learn more about all these week’s stories, make sure to check the description box below for all the links.
3:23 · Please remember to like, subscribe, share, and stay safe.
3:26 · I’m Rody for the Developer Show.
3:27 · Thanks for watching, and we’ll see you next week.
3:29 · [MUSIC PLAYING]