News for hackers. Stuff that matters.

GirkovArpa

Ask HN: How to OCR a PDF and preserve whitespace?

I have some rather large PDFs that need to be transcribed, but every service I try has some minor but deal-breaking flaw.Either they don't support PDFs this large (hundreds of pages), are just really bad at English OCR, or, most commonly, don't preserve whitespace correctly.The number one problem is whitespace when it comes to multiple columns (similar to newspapers). Either not putting any...

( Read More... | 17 comments | on HN )
Re:Why don't humans have tails? on Monday March 25, 2024 @06:56PM (#39822188)
This comment needs to be higher.
Re:Large Language Models' Emergent Abilities Are a Mirage on Monday March 25, 2024 @06:47PM (#39822104)
> You can ask them to reverse numbers or re-arrange words and they'll faceplant in the same way as soon as the input gets beyond a small threshold. Here surely there wouldn't be an issue with tokenization.

My guess is the training data contains many short pairs of forward and backward sequences, but none after a certain threshold length (due to how quickly the number of possible sequences grows with length). This would imply there's no actual reversing going on, and the LLM is instead using the training data as a lookup table.

This should work fantastic in theory, since differing vocabulary (not grammar) is the main factor that determines the difficulty of a new language. Putting off this primary obstacle so one can ease into it sounds genius to me. It also agrees with the method hyped by Steve Kaufman, where one should read and speak level-appropriate material.
Exactly. The proof they did not actually crack Bitcoin is the fact they said they did.
In the context of Condorcet's jury theorem, the percentages refer to the chance of voting for the correct outcome. Think of a legal trial and there is no ambiguity about the meaning of "50%" is.
According to Condorcet's jury theorem, a committee of 10 identical members may be smarter than a single member.
Re:Netlify just sent me a $104K bill for a simple static site on Tuesday February 27, 2024 @02:12PM (#39528291)
Yes. Back around 2020 they forgave me a $1000 bill for a side project I thought was running on a free tier.
Re:Show HN: Minimalist CSS Framework on Tuesday September 27, 2022 @07:42PM (#33002595)
I love it! Going to use this instead of Pure.CSS from now on.
Hello HN — This is a desktop app that accepts typed text and synthesizes a handwritten version of it, using recurrent neural networks in real time. The network is implemented as a library (included in the Releases section), written in 100% Rust. The GUI is based on the free Sciter.JS library, which supports SVG and the fancy controls. All this is an unofficial port of https://calligrapher.ai

By clicking the download button at the top-left, the output can be saved as SVG.

Re:Sweden is being shot up on Wednesday July 28, 2021 @01:57AM (#27980628)
> All the languages in Europe descended from the Proto-Indo-European language

Except for Basque!

Hey HN — This is a simple web app that allows you to draw an alphabet, type some text, and then have the text written in your own handwriting. I’m happy to hear feedback or answer questions about how it works!

P.S. Drawing the entire alphabet isn't needed, only the letters you need.

Compare to:

https://www.calligrapher.ai/

https://www.cs.toronto.edu/~graves/handwriting.html

Sciter is really great.
> you can't distribute the binaries yourself

You can :) You just have to include a copyright notice unless you purchase an exemption.

Re:Ask HN: Desktop Programming Language on Friday April 02, 2021 @04:42PM (#26675766)
I recommend Sciter.JS as a lightweight alternative to Electron. It's less than 10mb.

If you visit the repo here: https://github.com/c-smile/sciter-js-sdk

There's a database demo in samples/sqlite.

Re:Nuklear: A cross-platform GUI library in C on Tuesday February 23, 2021 @12:27AM (#26234095)
How does this compare to Sciter? I understand this is primarily to overlay UIs over full screen applications (like games), but Sciter has that capability as well. Particularly how easy is it to style things, considering Sciter allows CSS?

I did a CTRL+F for "sciter" and usually I find something, but this time, nope.

Re:Tauri: An Electron alternative written in Rust on Saturday February 20, 2021 @04:42PM (#26208247)
Would you be able to mention these apps? I would like to take a look.
Re:Tauri: An Electron alternative written in Rust on Friday February 19, 2021 @08:20PM (#26200417)
I am convinced that the sort of Web API compatibility so many people demand is only a few native libraries away. I would like to start experimenting in this area but I can only get rust-sciter DLLs working with "vanilla" Sciter and not Sciter.js, and I don't want to use other languages.
Re:Tauri: An Electron alternative written in Rust on Friday February 19, 2021 @08:15PM (#26200385)
I've performed a couple experiments remaking Electron apps using Sciter.

My first target was https://github.com/AkashRajpurohit/clipper, a neat little clipboard manager. The owner was gratious enough to officially list my project under the "Clones" section of his readme =D The result was a 6mb file, compared to the original 165mb Electron app.

The second attempt was https://github.com/girkovarpa/temps-lite, an aesthetically-pleasing weather app which was motivated largely by the fact that the original was broken and abandoned yet still had people who wanted to use it. According to the open issues and forks trying to resurrect it, anyway. The file size savings were similar to the former, and they start virtually instantly. Unlike the Electron apps which have a bit of delay and then a blank window before they finish starting up.

Compare the 32-bit Windows releases of Temps (42.6mb), made with Electron:

https://github.com/jackd248/temps/releases/tag/v0.7.0

... to the Temps -"lite" remake (4.49mb) using the Sciter engine:

https://github.com/GirkovArpa/temps-lite/releases/tag/v1.1.3

Temps-lite (and of course it's predecessor, Temps) is a desktop weather application for Windows, Mac, and Linux. Provided a free openweathermap.org API key, it displays live weather information for any location in a very attractive, sleek interface. It also shows the forecast for the next 4 days.

To duplicate the very cool rain, snow, and thunder effects, I basically just copied and pasted the original source code. For everything else I redid it in my own way, which probably did not result in very professional-looking code but hey! It works =) Minus some tiny bugs, which I've had help with from someone who's more familiar with Sciter programming.

The original Temps was built/packaged using Electron Builder and Electron Packager (I'm not sure of the difference, to be honest).

The equivalent for Sciter is Quark (quark.sciter.com), and that is what I used to package the HTML, CSS & TIScript files into a single convenient executable which is not only 10x smaller than the Electron version but also starts up way faster (virtually instantly), and has a relatively extremely light memory footprint.

All this without the slightest observable decrease in performance or speed.

One new thing I learned about Sciter while making this project is that it natively supports persisting its equivalent of local storage to a database file, similar to MongoDB.

If you think this is interesting then feel free to check out Clipper-lite, which is the same idea. Another Electron app remade using Sciter. Specifically, a clipboard manager:

https://github.com/GirkovArpa/clipper-lite

Re:Parisian Accent in 1912 on Thursday January 14, 2021 @08:26PM (#25785724)
push, p/eu/sh?

long, l/on/ng, l/an/ng?

Re:Kivy: Cross-Platform Python Framework for UI Development on Sunday December 20, 2020 @06:55PM (#25490761)
How does Kivy compare to Sciter?
I'm reading this while listening to music.
Re:The mythical “fast” web page on Wednesday December 09, 2020 @12:46PM (#25361325)
Like Google itself.
Re:Why I chose Electron.js for my side business on Friday November 20, 2020 @07:26PM (#25166779)
Sciter is missing from that list for some reason.
Re:Why I chose Electron.js for my side business on Friday November 20, 2020 @07:21PM (#25166752)
Sciter.JS is going to change that: https://github.com/c-smile/sciter-js-sdk

I cloned a 165mb Electron app in Sciter, and it was only 6mb:

https://github.com/GirkovArpa/clipper-sciter

Older »