Thanks for the correction and clarification! I just assumed from the open-r1 post that they gave everything aside from the training data.

Open source isn’t really applicable to LLM models IMO.

There is open weights (the model), and available training data, and other nuances.

They actually went a step further and provided a very thorough breakdown of the training process, which does mean others could similarly train models from scratch with their own training data. HuggingFace seems to be doing just that as well. https://huggingface.co/blog/open-r1

Edit: see the comment below by BakedCatboy for a more indepth explanation and correction of a misconception I’ve made

As a resident that still angers me.

I’m unsure if the original inventor is still alive for you to ask that question, but let us know the answer if you do get it.

https://en.m.wikipedia.org/wiki/Splayd

I just use PLA. PLA itself is good safe, but occasionally the additives aren’t, so I don’t use any for human related stuff. It’s also worth considering that the layered approach can allow for bacterial growth, so unless you treat it (e.g. epoxy seal it), you’ll need to wash it fairly frequently to curb buildup.

I wouldn’t say I’ve made back my investment on 3D printing in the past half a decade I’ve done it. But in terms of “prints for friends” like this one above I may be close. Plus there’s just something nice about going “I need a measuring cup for dog food” and printing one to the exact serving size.

Both should if your goal is to not have a reusable fingerprint (which for a privacy focus would be). Server should change more frequently since it has access to less information about the browser. Server based fingerprinting is fairly unreliable, client side uses Javascript to generate more bits of unique data.

The bottom result (the % certain one) is just a fuzzy match of similar fingerprints AFAICT.

The test is simply showing two fingerprints for your browser. One, the server fingerprint, is one that any tracker can see. The other, the client fingerprint, is what can be used if you have Javascript enabled.

Instead of inundating you with test results, this one is simple - check to see if your fingerprints change between browsing sessions. If they don’t change, that means you can be tracked. In which case you can mess with settings and try again.

So server code is your fingerprint based on what a server is able to see. This would be your fingerprint with JS disabled, essential. Client code is the JS generated fingerprint.

For the emojis I have no idea.

One of the points of Libre Wolf is to make you unique, but each session should be unique.

You can find some additional setting tweaks here: https://librewolf.net/docs/settings/

The “letterboxing” feature is an additional uniqueness feature you could consider enabling.

I’m particular you could check your result in this experiment: https://fpresearch.httpjames.space/

Try it in both normal and in a private tab, then record those results, reopen Libre Wolf, and try again.

The ad serving companies (Google) don’t care about what happens after the click (yet). As far as I’m aware no “handshake” process exists that would allow an advertiser to communicate with the as server and validate a click (such a process could be abused).

Most likely the advertiser would be using some form of client side analytics, so the click wouldn’t show up in their statistics, meaning the advertiser would see a huge discrepancy between the clicks they saw in the campaign and the clicks the ad server reports.

That’s what I thought at first, but the person who wrote the article is named Simon, and based on the context given in the article I’m assuming that was a test unit he had on his desk, but the planned implementation is in bathrooms.

Considering it only detects if someone in the bathroom is vaping and not who, disciplinary action just isn’t really possible with your typical school restroom.

Programming and self hosting the results when I was ~14 is what led me to a tech background. No university, but I’ve been working professionally in both IT and software for over a decade and self hosting even longer.

The article that user links is referring to GrapheneOS (and other OSS software) as not being “free software” - and they (GNU) delves into it more here.

Basically, GNU is saying software shouldn’t claim to be free and open source if they contain non free binaries / other non-free blobs.

The nuances between FOSS and OSS can be confusing. GrapheneOS is not claiming to be FOSS.

This behavior is because Firefox is using the MIME type of the file on the save as dialog.

Regarding your issue importing them - does renaming the ones in the ZIP to m4a not work? If the issue is mixed mp3s and m4as, you could use something like ffmpeg probe to rename them.

Or just skip it entirely, I use the Consent-O-Matic extension which has a surprising amount of features.

Brave has been off limits for me ever since I saw my QAnon nutjob father using it lol.