elephantum

That's the issue with Google, no specific user group is significant enough for Google to care

My biggest frustration with WebRTC was precisely captured in the article: even if you don't need p2p and your video source is the process on the same host with your browser, you have to dance around connection setup like you're on a different side of a planet

Making this case public should drop a shadow of doubt onto other Ukraine statements

Probably it would be wiser not to brag about this one

You do know, that you can create more than one table in SQLite and have references from one to another? Even recursive references work

I do not see direct statements that they will stop maintaining sources in open source.

We'll see :)

It looks like setting up a mirror and CI/CD on top of Github might work for some time. ghcr is free for public images

They write in the press release, that the sources remain under Apache 2 license, they just stop distributing prebuilt images for free.

Edit: As I see it's true.

Source code for OCI images: https://github.com/bitnami/containers/tree/main/bitnami

Charts: https://github.com/bitnami/charts/tree/main/bitnami

So, are they evil because they decided to stop sponsoring free network egress?

I tried to do brick sorting (because we have great detection and classification models at https://brickit.app/)

It turned out to be much more complex than I expected.

The biggest issue was grabbing. Typical approach for this type of task is to use vacuum suction actuator, but it does not work for Lego parts, because they have stubs and prevent suction from working.

Also there are issues of part separation.

We abandoned this idea, but I still hope that we can return and achieve something working some day.

Everything runs on device in tflite and it gives us some headaches, especially Android ecosystem.

We do not use anything fancy, detector is ssd-like, and classifier is either resnet or efficientnet depending on your device capability

The use case "let me reassemble all the parts that I have" is out of scope of Brickit app. It works explicitly with the contents of a single scan, that's why you see small ideas to build: it's just what fits into the parts you scanned.

But! We recognize the intent, and we have something in the works which will be released very soon, stay tuned!

This one actually works :)

Source: I'm responsible for ML development at Brickit

Anoop, first of all: thanks for your efforts, we switched to Bruno from Postman and are quite happy!

And a question: can you share your plans about Vscode extension? It seems to be broken for a while, do you have plans/capacity to revive it?

Interesting, is there a possibility of something like class-action where iPhone owners sue Apple for restricting access to Fortnite? I bet there should be a plenty of angry gamers.

I read this and cannot believe that I can optimize our $5-6K GCP egress bill to zero. Just wow.

There's a plugin for VSCode that brings draw.io editor for local .drawio files. IMO it's the best of both worlds: nice editor and git version control.

Also this plugin works fine in github.dev so it's like a better version of draw.io official site.

We tried Coroot recently in our production AWS cluster.

I must say it's handy. Cost analysis helped us shave ~20% of our EKS bill

Foxglove is my best argument to use ROS instead of custom solution in semi-robotics cases (several cameras/sensors + ML, but no actuators)

Just want to thank you for all the work you do! We are heavy users of sqlalchemy, our life would be much harder without it!

I have to mention, that what you described is not what ML guy is supposed to be doing.

Russia does that already country-wise.

Every receipt is sent to federal taxes office instantly, every cash register is connected to the internet.

How is it different from zeppelin?

Did it also handled similar load and feature complexity as well?

Fugue is a translation layer from SQL to underlying runtime: pandas, dask, spark.

Each of the runtimes, supported by Fugue, can be compared to Ray, but Fugue is a tool of a different kind.

no

The thing is, that it's all very time-sensitive. And I expect pressure was unreasonably high.

We have elections in Russia this weekend and Navalny app was used as a tool to coordinate protest voting. It is called "Smart Voting" by Navalny team and it is getting a lot of attention for upcomping elections.

If Apple and Google would just be a little bit slower and complied next Monday :)

Still there are more ways to coordinate Smart Voting. Google Docs with recommendations are published. There are youtube videos where Navalny team reads whole list aloud. There's a Telegram bot.

So I hope this action would just bring more attention to the matter.

Nice, are there any benchmarks?

Would be interesting to see how it compares to Postgres or LevelDB for read/write of exact values

And how it compares to Faiss/Annoy for KNN

Wow! Thanks for such a detailed answer. It's much clearer now.

Do I understand correctly that using XNNPACK and mobile acceleration is mutually exclusive? I.e. it's either XNNPACK or NNAPI/CoreML?

Should I consider XNNPACK for a modern mobile phone?

Thanks!