I know that Sans Display has been deprecated and tossed in the bin, but is Serif Display also deprecated, or just active-but-quiet? Noto Serif Display v2.009 is listed under Releases here, and the files are listed on the Dashboard page. ...

I'm not sure they were, though; that's the problem. The "display" version was not exactly optical size and not exactly contrast, and was a different implementation between serif and sans. It didn't really fit into any of the models of font variation that we support elsewhere. So they need re-thinking, harmonising, and re-drawing.

You can consider them to be deprecated.

Noto Latin, Greek, Cyrillic. Contribute to notofonts/latin-greek-cyrillic development by creating an account on GitHub.

Serif does not have a full build. Sans has a full build for messy historic reasons.

Once upon a time when Google Docs/Workspace was young, they requested
Noto Sans but they also wanted it to work seamlessly with documents in
Hindi without people needing to change fonts, so they asked for the
version of Noto Sans in Google Fonts to contain a Devanagari glyphset.
Now Workspace's support for other languages is better but once you've
released a font with some glyphs in it, documents will break if you
take them out, so it's got to stay. So for the exceptional case of Noto
Sans, "full"/"googlefonts" builds mean "we added a Devanagari core".

Markdown extraction is just the first step. For documents to be effectively used in RAG pipelines, they must be split into smaller, semantically meaningful chunks.

Recent studies have shown that using large language models (LLMs) for this task can outperform other strategies in terms of retrieval accuracy. This intuitively makes sense - LLMs excel at understanding context and identifying natural boundaries in text, making them well-suited for generating semantically meaningful chunks.

The problem? Cost. Until now, LLM-based chunking has been prohibitively expensive. With Gemini Flash 2.0, however, the game changes again - it's pricing makes it feasible to use it to chunk documents at scale.

https://news.ycombinator.com/item?id=42952605

(disclaimer I am CEO of llamaindex, which includes LlamaParse)
Nice article! We're actively benchmarking Gemini 2.0 right now and if the results are as good as implied by this article, heck we'll adapt and improve upon it. Our goal (and in fact the reason our parser works so well) is to always use and stay on top of the latest SOTA models and tech :) - we blend LLM/VLM tech with best-in-class heuristic techniques.

Some quick notes: 1. I'm glad that LlamaParse is mentioned in the article, but it's not mentioned in the performance benchmarks. I'm pretty confident that our most accurate modes are at the top of the table benchmark - our stuff is pretty good.

2. There's a long tail of issues beyond just tables - this includes fonts, headers/footers, ability to recognize charts/images/form fields, and as other posters said, the ability to have fine-grained bounding boxes on the source elements. We've optimized our parser to tackle all of these modes, and we need proper benchmarks for that.

3. DIY'ing your own pipeline to run a VLM at scale to parse docs is surprisingly challenging. You need to orchestrate a robust system that can screenshot a bunch of pages at the right resolution (which can be quite slow), tune the prompts, and make sure you're obeying rate limits + can retry on failure.

paul buchheit on the advent of gmail, insurgents vs. gatekeepers, the future of san francisco, ai,

An insurgent is never going to be successful working for gatekeepers, because from the gatekeeper perspective, the person just lacks maturity. They're doing these things that are obviously risky. That's the nature of throwing bombs: you don't exactly know what's going to happen.

Famously, we had one line of JavaScript in the entire product, which was when you go to the homepage, there was one snippet of JavaScript that would put the focus in the search box. That was the only JavaScript at Google. So Google had this kind of anti-JavaScript thing, that was also partially technical snobbery. One of the senior technical people said, “You can never scale anything in JavaScript, it’s basically just shit, the project will just turn to garbage.” And actually Eric Schmidt even said, “Oh yeah, my friend at some other big company tried this, it doesn't work.” There was a lot of ‘This has been done before, you just don't know why it can't work.’

Gatekeepers are one hundred percent anchored to stopping bad things from happening, and they have no concept that when you stop bad things from happening, you are inherently stopping good things from happening as well. You can't ever deliver something that's 100 percent good. If you deliver 80 percent good, that's pretty good. But if you try to go for 100% — if you try to be perfect — what you get is nothing. Innovation is inherently not clean.

A description of advanced tips and tricks for effective Internet research of papers/books, with real-world examples.

used book search engines: Google Books/find-more-books.com
it’s worth searching the seller to see if they have their own website, which is potentially much cheaper.

Search engines in general have realized that it's more profitable to show you irrelevant results than to show you nothing. Furthermore, they've realized it's more profitable to show you irrelevant results laden with their ads than show you highly relevant results from ad-free sites.
Perverse incentives at work!

This is precisely what happened. When google merged with doubleclick.net the new company should have been named doubleclick.net and not google. The old google ceased to exist at that point and was swallowed by an advertising company.

phind.com
Kagi (don’t use it personally)

Material Symbols are our newest icons consolidating over 3,036 glyphs in a single font file with a wide range of design variants. Symbols are available in three styles and four adjustable variable font styles (fill, weight, grade, and optical size).

Colleagues, are there plans to combine the .glyphs sources for Noto Serif & Noto Serif Display into one source? And the same for Noto Sans and Noto Sans Display?

simoncozens commented yesterday
We decided to do away with the "Display" versions, as they were ill-defined. (Not really opsz, not really contrast, and not consistent between Sans/Serif.) Maybe we'll bring them back one day! (Probably not.)

https://github.com/notofonts/latin-greek-cyrillic/issues/436#issuecomment-1577379159
The "display" version was not exactly optical size and not exactly contrast, and was a different implementation between serif and sans. It didn't really fit into any of the models of font variation that we support elsewhere. So they need re-thinking, harmonising, and re-drawing.

Type in 'breathing exercise'

Barlow

https://fonts.google.com/specimen/Barlow?query=barlow#glyphs

Barlow is a slightly rounded, low-contrast, grotesk type family. Drawing from the visual style of the California public, Barlow shares qualities with the state's car plates, highway signs, busses, and trains.

This is the Normal family, which is part of the superfamily along with Semi Condensed and Condensed, each with 9 weights in Roman and Italic.

Encode Sans

https://fonts.google.com/specimen/Encode+Sans?query=encode+sans#about

Encode Sans is a versatile workhorse sans-serif superfamily ready for all kinds of typographic challenges, offering a unique blend of warmth and practicality. Its humanist aspects of simple letterforms and open apertures keep it crisp and legible, while its geometric approach of rounded letters with partially-straightened sides delivers a friendly but precise tone.

It includes 5 widths from Condensed to Expanded, each with 9 weights from Light to Black. To simplify the use of smallcaps in word processors, there are also small-cap versions of each family.

Cabin

https://fonts.google.com/specimen/Cabin?query=cabin

Cabin is a humanist sans inspired by Edward Johnston's and Eric Gill's typefaces, with a touch of modernism. Cabin incorporates modern proportions, optical adjustments, and some elements of the geometric sans. It remains true to its roots, but has its own personality.

The Cabin font family comes in two variable fonts, roman and true italic, with a Weight range from Regular to Bold, and a Width range from normal to Condensed. The stroke contrast is almost monolinear, although top and bottom curves are slightly thinned. Counters of the b, g, p and q are rounded, and all are optically adjusted.

Sarabun

https://fonts.google.com/specimen/Sarabun

Sarabun is an open source multi-script webfont that supports both Latin and Thai. It is the "TH Sarabun New" font

Prompt

https://fonts.google.com/specimen/Prompt

Prompt in Thai means “ready,” the same as in English. Prompt is a loopless Thai and sans Latin typeface. The simple and geometric Latin was developed to work harmoniously with the loopless Thai that has wide proportions and airy negative space. It is suitable for both web and print usage, such as magazines, newspapers, and posters.

A Hassle-Free Way to Self-Host Google Fonts. Get eot, ttf, svg, woff and woff2 files + CSS snippets!

The best Google Fonts can add personality and functionality to your website. You can even use Google Fonts with UXPin prototypes.

There are over 950 Google typefaces available. Probably, you don't know the majority of them. They vary in the number of variants and subsets.

The idea to tag Google Fonts born from our need to use less popular typefaces, with a strong identity.

One of the biggest challenges to quitting is the fact that most alternatives, particularly those in the open source of privacy space, are really not user friendly. I’m not a techie. I have a website, understand how to manage Wordpress, and can do some basic troubleshooting, but I can’t use Command Line or do anything that requires coding.

As a web designer, you’re probably well aware that Great fonts are essential to great design.
This article is all about fonts. We have collected 14 unbelievable Google Web Fonts that makes your heading or headline of your website attaractive.
The fact is that the headlines are more important than you may think. and Many designers also agree this point.

Nice 'code' css style