mseri

Ok, done: https://github.com/mseri/zunzuncito My main focus is systems with very low ram. On my M1 with 8Gb or RAM llama.cpp cannot run gemma4 26b, but this implementation works fine with 5-6 tok/s.

I also replaced my crap implementation of the oai server following yours, I hope it's all right. I did add a mention to your project and this fact in the README. It is a lot more barebones than what you have, but I have to admit that it works really nicely for me

Nice!

I had done the exact same with gemma4 26b, both for my Intel laptop and for my M1 with 8Gb RAM (with also q4 and turboquant). I don’t use it much since there are dumber but way faster models to run, but I should clean up the code and make it available

Finally! Great to hear. I had the exact same experience, but gave up at the moment of ID verification… too much hassle indeed

As much as I agree with the need for digital independence and the fact that universities (and governments) in Europe are over reliant on US tech, it is not as simple as you describe.

There is a lot more happening in the administrative and infrastructural side of things in most universities that one barely observes as student. So every change needs to take also that into account, the management and maintenance of services and infrastructures that must reliably support thousands of users, with relatively strict privacy and security standards, and their migration.

See also https://news.ycombinator.com/item?id=46080495

Thanks indeed!

Thanks for the link. I had missed the other two submissions.

If any admin is around, they should probably be merged. This is the other one: https://news.ycombinator.com/item?id=46055863

Google has a great aid to reduce the attack surface: https://github.com/google-research/arxiv-latex-cleaner

I just want to say thanks to him!

And very poor article by Politico

META did pirate basically all books in Anna’s archive but if I remember correctly they just whispered a a cried sorry and it ended up as that. Why are they also not asked to pay?

True, but we should also remember that some services like the fast responses and the image generations (may?) run in US data centres also for Mistral. So that part of the data, in principle, may end up in the ends of other extra European countries.

This said, I am really supportive of Mistral, like their work, and hope that they will get more recognition and more EU-centric institutional support.

Sounds all cool and interesting, however:

> By submitting User Submissions through the Services, you hereby do and shall grant Inception a worldwide, non-exclusive, perpetual, royalty-free, fully paid, sublicensable and transferable license to use, edit, modify, truncate, aggregate, reproduce, distribute, prepare derivative works of, display, perform, and otherwise fully exploit the User Submissions in connection with this site, the Services and our (and our successors’ and assigns’) businesses, including without limitation for promoting and redistributing part or all of this site or the Services (and derivative works thereof) in any media formats and through any media channels (including, without limitation, third party websites and feeds), and including after your termination of your account or the Services. For clarity, Inception may use User Submissions to train artificial intelligence models. (However, we will not train models using submissions from users accessing our Services via OpenRouter.)

Some more details here: https://arstechnica.com/tech-policy/2025/06/openai-says-cour...

And here are the links to the court irders and responses if you are curious: https://social.wildeboer.net/@jwildeboer/114530814476876129

Jim Portegies (TU/e, Netherlands) and Jelle Wemmenhobe have done a lot of research on this, using their “waterproof” (controlled natural language compiled to coa) to test this directly in class. The results are very interesting, and indeed actively messing around is still a very important part of the learning experience, but you can see at least some benefits in also having a theorem prover to check if your proofs are correct.

What I was surprised is that the students learn some patterns of proof properly, but only if you make sure that they are explicitly exposed by the proof assistant (so the more automation the less learning also in this case).

You can find a lot of the work summarized in Jelle’s PhD thesis at https://research.tue.nl/nl/publications/waterproof-transform...

It is a bit annoying that the article does not link any relevant research. There is a wikipedia page on the topic (https://en.wikipedia.org/wiki/Coffee_ring_effect), but afaik it is an interesting problem in many different contexts, for example in inkjet printing (one can find plenty of articles there as well).

It reliably fails also basic real analysis proofs, but I think this is not too surprising since those require a mix of logic and computation that is likely hard to just infer from statistical likelihood of tokens

You can choose the quantization by appending the right tag to the model name, but they don't support other more advanced useful features (e.g. you need a special flag to enable flash attention and you cannot use KV cache quantization for large contexts).

Both I’d say. More recently also online calls/conferences/seminars, and (way more rarely but it happens) on twitter/mastodon/…

Do you remember the one they released when they changed name to linspire: https://www.youtube.com/watch?v=IIYtKHnU4mQ memorable

Without paywall at https://archive.ph/8aj7W

Reason and ReScript are two different projects. You can still use Reason to compile natively and to JavaScript via js_of_ocaml.

The book is open access (https://linear.axler.net/), I am going to quote directly the author in the preface:

<< all linear algebra books use determinants to prove that every linear operator on a finite-dimensional complex vector space has an eigenvalue. Determinants are difficult, nonintuitive, and often defined without motivation. To prove the theorem about existence of eigenvalues on complex vector spaces, most books must define determinants, prove that a linear map is not invertible if and only if its determinant equals 0, and then define the characteristic polynomial. This tortuous (torturous?) path gives students little feeling for why eigenvalues exist. In contrast, the simple determinant-free proofs presented here (for example, see 5.21) offer more insight. Once determinants have been banished to the end of the book, a new route opens to the main goal of linear algebra— understanding the structure of linear operators.>>

If you like mathematics, it is actually a pretty nice book.

Seems to be up and running fine. Gitea seems to be quite robust and reliable! I hope you have regular backups, I had a number of unfortunate IO corruptions with the PIs.

Thanks for the open sourcing

Yes but it is not really relevant for that book. OCaml has an incredible backward compatibility history, and the linked book is about learning the language itself, not its ecosystem.

Things are different for RWO since it relies also on a number of extra tools and libraries which, indeed, change over time (sometimes also substantially)

There are a number of interesting projects already doing this, either doing it manually via the c FFI or using projects like ocaml-rs [1] or ocaml-interop [2] for example

[1]: https://github.com/zshipko/ocaml-rs [2]: https://github.com/tezedge/ocaml-interop

They do, they are developing drom[1], a wrapper over opam and dune to provide a cargo-like experience.

[1]: https://ocamlpro.github.io/drom/

What’s wrong with opam? Since version 2 it also supports local switches and locking, and the opam repository has become much more stable than in the past. Full disclosure: I am biased here, since I am one of the opam-repository maintainers and I am quite happy with opam itself.

Yes, have a look at [0] and maybe also the one in [1]

[0]: https://github.com/ocaml-bench/sandmark [1]: https://github.com/ocaml-multicore/retro-httpaf-bench/pull/1...

I had the same experience. I installed the RC and it has been working surprisingly smoothly on my intel macbook pro (15’’ 2018)

You can see here what will work and what not: https://github.com/onivim/oni2/issues/1058

I see, I don’t know but its license is pretty clear in saying that you cannot use it outside of Microsoft stuff: “You may install and use any number of copies of the software to use solely with Microsoft Visual Studio family of products.”

I don’t know. What is live share?