HackerTrans
トップ新着トレンドコメント過去質問紹介求人

jeffjeffbear

no profile record

投稿

The Pacific Northwest Tree Octopus

zapatopi.net
2 ポイント·投稿者 jeffjeffbear·5 か月前·0 コメント

Frankenwine: Multiple Personas in a Wine Process

nullprogram.com
1 ポイント·投稿者 jeffjeffbear·6 か月前·1 コメント

Web Browser Engineering

browser.engineering
1 ポイント·投稿者 jeffjeffbear·6 か月前·0 コメント

Understanding Encoder and Decoder LLMs

magazine.sebastianraschka.com
1 ポイント·投稿者 jeffjeffbear·7 か月前·0 コメント

コメント

jeffjeffbear
·5 か月前·議論
Could be good as a modern version of https://zapatopi.net/treeoctopus/
jeffjeffbear
·6 か月前·議論
> just to be clear, semiseparate in this context means H = D + CC', where D is block diagonal and C is tall & skinny?

Not quite, it means any submatrix taken from the upper(lower) part of the matrix has some low rank. Like a matrix is {3,4}-semiseperable if any sub matrix taken from the lower triangular part has at most rank 3 and any submatrix taken from the upper triangular part has at most rank 4.

The inverse of an upper bidiagonal matrix is {0,1}-semiseperable.

There are a lot of fast algorithms if you know a matrix is semiseperable.

edit: link https://people.cs.kuleuven.be/~raf.vandebril/homepage/public...
jeffjeffbear
·6 か月前·議論
I haven't looked into it in years, but would the inverse of a block bi-diagonal matrix have some semiseperable structure? Maybe that would be good to look into?
jeffjeffbear
·6 か月前·議論
> Haven't used Linux in forever, but middle-click to paste was like the one thing that consistently worked everywhere.

That's because it was an X11 thing, and everyone used X11.
jeffjeffbear
·6 か月前·議論
That is also the one good thing about Window's commandline, you use right click there to copy and paste which is nice. The rest sucks.
jeffjeffbear
·6 か月前·議論
I don't really love this solution since it runs into all the usual linked list issues, and is only 'allocation free' in the sense that the pointers are allocated with the structure if doing the intrusive thing they are talking about. Using the std::vector of pointers approach isn't going to be using crazily more memory.

Myself, I like to just allocate a too big block and shove everything into that then deal with indicies into that array if I care about performance. you can even flatten the tree in a way to get better locality if you care about that.
jeffjeffbear
·6 か月前·議論
I would really like to see more testing with a deeper hierarchy and alpha and beta nonzero.
jeffjeffbear
·7 か月前·議論
> I think it really depends on what kind of anime you’re talking about

Does it? If I draw a naked stick figure with boobs and say it is 14, is that morally wrong? At what point should a person care? Their point is that a drawing doesn't hurt people right?
jeffjeffbear
·7 か月前·議論
They have some more details at https://github.com/DGoettlich/history-llms/blob/main/ranke-4...

Basically using GPT-5 and being careful
jeffjeffbear
·7 か月前·議論
I just don't like having to send HTML and have the backend deal with what are really frontend problems. Sending JSON is great since you can serialize most reasonable data types into it and then the backend has no responsibility for how it is rendered which helps for having mobile apps use the same backend as the website. Sending HTML just seems nuts since if you change your design you would have to change the backend too.
jeffjeffbear
·7 か月前·議論
> https://huggingface.co/google/t5gemma-2-1b-1b

From here it looks like it still is long context and multimodal though?

>Inputs and outputs Input:

Text string, such as a question, a prompt, or a document to be summarized

Images, normalized to 896 x 896 resolution and encoded to 256 tokens each

Total input context of 128K tokens Output:

Generated text in response to the input, such as an answer to a question, analysis of image content, or a summary of a document

Total output context up to 32K tokens
jeffjeffbear
·7 か月前·議論
Isn't finetuning the point of the T5 style models, since they perform better for smaller parameter counts?
jeffjeffbear
·7 か月前·議論
You have control over what displays on a page with a discord.com domain, you could manipulate the dom to have a login or something else and have it pass the data to your servers. A user would just see a link from discord.com
jeffjeffbear
·7 か月前·議論
I don't think the fee to get access to the standard is generating much income for anyone. Most of what your talking about seems to be money made from licensing of the technology, right?
jeffjeffbear
·7 か月前·議論
When I was younger I thought of replacing most of the OS with a browser since that is how I used it. but this is weird and not in a good way. Maybe using Firefox would feel better.
jeffjeffbear
·7 か月前·議論
I run a math circle in my area, and use forgejo with kids for thier solutions in Latex and python, it works great for me, and is super easy to enforce logins and reset passwords.
jeffjeffbear
·7 か月前·議論
I have seen this particular idea come up a lot lately.

Personaly I think Intel's early investment in linux had a lot to do with it. They also sold a compiler and marketed to labs and such which bought chips. So linux compatibility meant a lot to decision makers.

AMD the underdog went more in on Linux compat than NVIDIA. Which may have been a business decsion.

I dunno, maybe the GPL effect was more a market share thing with developers than a copyleft thing.

Nota Bene: I do love copyleft and license all my own projects AGPL
jeffjeffbear
·7 か月前·議論
I would still use it but NVIDIA abandoned it, so I can't. I wish they still had some support for modern CUDA.
jeffjeffbear
·8 か月前·議論
I don't mean to be mean, or anything, but aside from educating yourself this isn't super useful for others, if it could run any Qwen3 model or something it might be more interesting. Maybe have some explanation of what you are wanting from this project?
jeffjeffbear
·8 か月前·議論
Doesn't seem very bad, for any of the Zig code I looked at, it would maybe have been nicer to try starting from something like GGML and building on top so it was more extensible. Personally, aside from implementing all of BLAS in Zig, I would have just used OpenBLAS.