iamjackg

That would have been really impressive in 2016!

I found my narrator by being married to her, and all the animation was done in After Effects.

Absolutely agree. The main downside of that pattern is that it doesn't work with jobs included from other projects in GitLab CI, since the job runs in the context of the project that imported it and therefore can't find the script in its original repo. Huge bummer.

Oh, that's really cool. I was trying to solve this from a different perspective a while ago: I wanted to add some pre-processing that would take a "normal" shell script and render it to the `script` part of the corresponding job at build time, the advantage being that you still have everything self-contained in the gitlab-ci job.

I stopped working on it because dealing with shell shenanigans in the GitLab CI runner environment is such a miserable experience that we're in the process of moving all our jobs to python scripts.

Thank you so much! It's all done in After Effects, with some Photoshop work for more complex graphics. The voice over was kindly provided by my wife.

Briefly, but I realized that the creative satisfaction of the whole thing came from being able to do everything myself, so giving that up would have made it pointless to me. I feel bad about it because it's definitely a bit of a narcissistic take (praise me, for I can do it all!), but the truth is that I was also discouraged by the mistakes I made in the research process, and how mean some of the commenters who pointed them out were.

I started to get a lot of anxiety about the whole process, and realized that I would have had to put even more effort into it ok top of my full time job, which further cemented my choice to just run away from it. I periodically feel really bad about it. Hopefully some day I'll win the lottery and I'll be able to make another episode. I actually have the script for episode 3 all done: I wrote it on parental leave, but that also didn't last forever!

Yeah! Search for "zombie mode" on this page: https://gbdev.gg8.se/wiki/articles/Gameboy_sound_hardware

Oops! I just added it.

A long, long time ago I started a YouTube series[0] on the internal workings of the Game Boy during a brief break from work. I eventually started working again and animating the episodes was taking way too long, so I had to stop, but I've kept orbiting the world of Game Boy reverse engineering.

The amount of dedication, skill, and passion that goes into it on a daily basis is mindblowing, and people keep discovering hardware quirks and bugs to this very day! I believe it wasn't that long ago that people discovered a glitch in the audio subsystem that gave you more control over note envelopes, and it was promptly added to LSDj (a tracker for the Game Boy that was first released in 2000) by the original author, who's still maintaining it to this day.

0: https://www.youtube.com/watch?v=RZUDEaLa5Nw

EDIT: oops, forgot the link!

If a business can't handle its own scale without negatively affecting its customers, it should probably stop growing. It's the same issue we see with Google accounts being seemingly randomly terminated.

Since that probably won't happen, heuristic usage should at least come with penalties attached, otherwise the incentives are lopsided. If an airline's overbooking heuristics fail and get you bumped, you either get put on another flight and/or receive financial compensation. If an insurance company's "heuristics" fail and deny a legitimate claim, there should be a penalty. If Google terminates your account because of a mistake, they should pay a fine. They shouldn't be allowed to have their cake and eat it too.

> Halftime D&B

Isn't that Drumstep? If not, I'd love some examples because that sounds very much up my alley.

> Why does Pendulum (a band) get a node on here? There is no such genre as Pendulum.

The guide actually explains that if you click on the dot:

"Pendulum is not the name of this genre. It's technically called Nu Jumpup or Nu Skool Jumpup but I call it Pendulum because Pendulum is more than just an artist or a genre.

It is an event.

It is one of those events that is so monumental that when people talk about history, they unconsciously draw a line between stuff that happened before the event and stuff that happened after (ie: "we are living in a post X world", where X is something earth-shattering like the atomic bomb or 9/11). We are witness to one of these life-changing events in Drum n Bass. Welcome to the Pendulum Age."

I'm not sure I would say your $14 doesn't go to the creators: they get paid per minute watched when the user has Premium. A while ago Lon Seidman published a video discussing his YouTube earnings, and he mentioned that Premium is actually much more remunerative for creators compared to traditional ads.

I used Opera Mini for many, many years on a variety of phones, from "dumb" feature phones to my N-Gage. What an incredible product it was! Thank you for allowing me to read cool stuff in Wikipedia while I was bored in school. It made things a lot more bearable.

I worked on a pull request to implement it in Terminator a long time ago, but eventually gave up. Control mode was contributed to tmux by the author of iTerm2, and if you explore the iTerm2 codebase you can definitely see that they're very tied together in terms of implementation choices and general design. It's not that simple.

In order to have two-way synchronization, your terminal needs to not only have an architecture where you can add a background thread that can control the "main" terminal interface, but also be flexible enough to properly size all windows and panes, and match font sizing. If your terminal supports different font sizes for every pane/split, those have to be turned off in tmux mode. There's all kinds of corner cases.

I'm sorry, I should have been clearer -- I meant that I wish this would be a locally self-hostable observability solution for OpenAI-compatible APIs in general, not something that can only be deployed to Azure and is tied specifically to Azure's OpenAI service.

For example, Helicone[0] and llm.report[1] work as generic proxies for the OpenAI API and you can deploy them anywhere.

0: https://www.helicone.ai/ 1: https://llm.report/

This looks great, although the Azure requirement makes it a no-go for me. I just keep wishing for a simple solution that is easy to self-host locally. After a lot of work I managed to get an instance of Helicone running, but it's made of 6 or 7 different services, each with a pretty large footprint.

I feel like there's definitely room for a single-binary tool made in Go with a small sqlite database and in-memory caching.

For those who don't know, it's a reference to Technology Connections[0], another incredible YouTube channel. Alec, the host, is very fond of that bit[1].

[0] https://youtube.com/@TechnologyConnections [1] https://www.youtube.com/watch?v=CZs-YcmxyUw

Reducing the precision of the weights from high precision floating points to either lower precision floats or even integers. You'd think it would greatly reduce the performance of a model, but in most cases the decline in quality is extremely tolerable compared to the reduction in memory/processing requirements.

For some reason the Max field in the picture says 42.4°, but if you look at the center target you'll see that it says 88.7°. I missed it too at first and was equally confused.

Very cool project! I'm mainly drawn to this because it seems easier to set up for self-hosting than Helicone[0], which is doing the exact same thing.

[0]: https://github.com/Helicone/helicone

EDIT: I take it back :) The self-hosted instructions don't really work: they assume you have node, yarn, and I guess prisma already installed?

Do you have a source on that being the reason? I'm extremely curious. The amount of work it takes to keysound a track (and figure out hot to not make it awkward at lower difficulty levels) seems like a much more reasonable reason not to do it, especially since there is some evidence[0] that Konami and Activision might have struck a deal for patent access, and even ended up collaborating to manufacture and distribute Guitar Hero Arcade.

[0]: https://www.gamedeveloper.com/pc/exclusive-exploring-i-guita...

That's worse than many Disney park attractions. They could take a lesson or two from their operations folks.

Looks great! Very excited to try this out. At first glance I have one concern and one suggestion:

Concern: you mention it supports the actual ChatGPT Web API -- isn't this against their terms of service? I'd be wary of publishing something that hits internal APIs.

Suggestion: I would rename what you call the "ChatGPT" (non-web) part of the backend to "OpenAI API" instead. There is already enough confusion in the world between ChatGPT and API access to the underlying GPT models.

I'm so confused. It seems like a joke (DankGPT and mentions of GPT5) but then it actually works. Is it just a tiny wrapper on top of langchain meant to poke fun at all the thin API-wrapper startups?

I had a great run in with BGR a few years ago when I found out that some production batches of a Dell monitor model had BGR panels instead of RGB. I also had trouble getting other people to see what I was talking about.

Ended up writing a blog post about it: https://blog.jackgaino.dev/is-this-a-bgr-i-see-before-me

> Isn't being freed from slavery the same as being freshly immigrated with 0$?

Well, if you're immigrating, it isn't <country you're moving to>'s fault you have no money.

If you've been enslaved, it's very much <country you were enslaved by>'s fault you have no money.

Type II fun: https://www.rei.com/blog/climb/fun-scale

I'm curious why you consider going to Disney faux-traveling. It's an opinion I hear often (and I once held) but everybody seems to get there for different reasons.

It's not quite to that level, but they carefully study the acoustics and position all speakers so the transition is gradual and not jarring. They might also choose songs that have elements of both lands to use in areas close to the transition point.

Semi-related: if you go on the Disneyland version of Pirates of the Caribbean, there is a short section of caves right between the first and second drop. The section only lasts maybe 10/15 seconds, but it's where you hear A Pirate's Life for Me for the first time in the entire ride as a kind of foreshadowing of what you will see later. Well, as soon as you go down the second drop, which is not that steep or long, the music is completely gone, and all you can hear is the atmospheric ambient music for the "main" cave section. I always found that incredibly fascinating, since you can turn around and see the previous area, but you cannot hear the music at all.

EDIT: there is actually something that is closer to iMuse, at least the way it was implemented in the Special Edition. Their parade routes are divided into multiple zones, and the soundtrack is equally divided in sections for each float. They use a positioning system to determine where each float is, and feed the correct section of the soundtrack to that speaker zone. It works incredibly well, and you don't notice it at all while you're there.

Listen to the soundtrack for the Paint the Night parade for an example: https://www.youtube.com/watch?v=SoIpk9F_Okk&pp=ygUhcGFpbnQgd...

Each section is basically the same length, the whole thing is (mostly) in the same exact key, of course it's the same BPM across the board, and each section has an intro/outro that can act as transition points to mix into the next section. They can also, of course, loop ad libitum if the parade were to stop for a few minutes. Really cool stuff!

Nintendo actually spent a considerable amount of time balancing exactly that metric for Zelda Breath of the Wild, according to a presentation they did in 2017. They replicated some Osaka landmarks in the engine to gauge how to space landmarks in the final game.

Here is one of the few remaining traces of this, since they asked for all conference slides to be taken down a few years ago: https://twitter.com/pixelpar/status/1286425979615936513

OpenAI released this earlier this month: https://openai.com/blog/function-calling-and-other-api-updat...