I think we're finally getting to see a glimpse of the future. People and LLMs, working together. (And doing it really well.)
It's pretty exciting.
282 karma · joined October 21, 2023
It's pretty exciting.
Like a full search engine that can visit pages on your behalf. Is anyone building this?
I realize it's far from a best practice and even explicitly stated as a bad idea somewhere in the SVG spec, but the idea of a document editor where you can individually position each and every character and make detailed, individual glyphs natively without loading fonts is interesting to me.
Are there any examples of this? Or perhaps different (better) approaches to a document editor with the advantages I said above?
Glad to see it reposted every now and then. Makes me nostalgic.
Sorry for the inconvenience.
Maybe give it another try? I'm on Safari as well.
Sucks that you even have to consider this though.
People (at least on HN) seem to be in agreement the Europe is too regulatory and bureaucratic, so it feels fair to question the practicality of any American initiatives, as we do for European ones.
What does this document practically enact today? Is there any actual money allocated? Deregulation seems to be a theme, so are there any examples of regulations which have been cleansed already? How about planning? This document is full of directives and the names of federal agencies which plan to direct, so what are the actual results of said plans that we can see today and in the coming years?
Found this clip while reading up on Apple's history, and thought it was interesting.
I guess this game is more representative than we'd like to think.
Others have made similar projects, but mine was made specifically for my use-case - I wanted to read the essays in my RSS reader, not by clicking a link.
I also wanted accurate, sorted dates; and a readily available hosted version. So I made this.
Might be useful to someone. Merry Christmas.
"I can't help myself" is not a valid excuse. If you seriously cannot bother to phrase things less dismissively, then you shouldn't comment in the first place.
One of the best guidelines established for HN, is that you should always be kind. It's corny and obvious, and brings to mind the over-said platitude my mom, and a million other moms, used to say: "if you don't have anything nice to say, don't say anything at all."
Your concession was admirable, but your explanation leads me to think that you misunderstand the role you play in the comments. You are not supposed to be a reaction bot; HN is not the journal for your unfiltered thoughts and opinions.
Despite how easy it would be, you cannot and must not simply write replies. Absolutely everything (yes, everything) written here should assume the best, and be in good faith. Authors and the community deserve that much.
This goes for other sites as well, but especially for a community that strives for intellectual growth, like Hacker News.
Apologies if I sounded harsh.
http://github.com/leftmove/facebook.js
I was working with the Facebook API for another project, but I was surprised that a website as popular as Facebook had exclusively old API wrappers for it, that all required too much setup, and were cumbersome to write code with.
I wanted a better approach using the standards I have come to appreciate with modern API wrappers, and so I decided to start work on a more intuitive, faster approach for the Facebook API.
My initial launch goal is to allow for programmatic authentication, posting, and commenting, all with one setup command through a CLI.
I don't even follow finance - I found this video by random - but Damodaran is clearly a great teacher.
I wasn't on the internet at that time, but apparently it was an early internet web hosting service.
Google seems to be paying tribute to it by changing the font of the entire page to Comic Sans; I assume this is to poke fun at the bare-bones websites of the old days.
Kind of cool.
I built Cria, a Python library that allows you to run LLMs programmatically through Python. Cria is designed so there is as little friction as possible — getting started takes just five lines of code.
I created this library because I was using OpenAI in my project, and kept running into rate limits. With local LLMs getting better and better, I sought to switch, but found command line configurations to be limited. Running and configuring LLMs can be trivial, but programs like ollama make it easier. The only problem I found with ollama though, was the lack of features in its Python client.
To counteract this, I built Cria — a program that is concise, efficient, and most importantly programmable. The name is based off of Meta's llama, as baby llamas are called crias. Cria solves pain points of ollama - it manages the ollama server for you, saves your message history, streams responses by default, and allows easier management of multiple LLMs at once.
While this isn't my first open source project, it's my first Python library. The code may be a little rough around the edges, so I would appreciate any and all suggestions.
Thank you!
This is likely due to the fact that the database is huge, and providing that data on demand is very resource intensive- especially when there are forty different people sending many requests a second.
If you don't want to compromise on data though, look into spending a little bit more time/money on infrastructure. I wish I had deployed the project on Kubernetes, instead of what I ended up doing.
This project was a way to use my web development experience for a good cause, and to learn a lot along the way. I hope to improve the project thouroughly though user suggestion, then maybe hand it off to other people. Like I said in my post, my main goal is to one day get into start-ups, so I can create good in the world through them. This is hopefully the first step in that journey.
Thanks for your kind words.
Thanks for informing me, and for the kind words, I'll make sure to avoid using the word company from now on.
The main feature I added because of WhaleWisdom was the data download. For any filer you can download all data in CSV or JSON, as on WhaleWisdom that's unavailable. I also plan to add a bunch of features those sites have eventually.
The problem with making a static site is that there are over 800,000 SEC filers, so it would be impossible to query all of them and store it.
I hadn't expected so much traffic, so I really have no clue how to handle this without excessive cloud charges. The best I've done so far is to look into free hosting for open-source projeccts and add a donation link to the homepage.
I have been tearing my hair out for the last hour trying to get my Always Free Oracle instance running again.
Other than the search database which needs to be available on demand, the rest of the filers are queried and analyzed on demand. This is required because some filers get really, really big. Blackrock Inc alone is about a 30MB file.
As for contribution, I am definitely looking for it. I have not been in the open-source field for long so I don't exactly know how to get it, but I would highly appreciate it if anyone could help.
Contribution would be especially helpful since I was still learning a lot of the technologies I used while I built this project, and the code is prone to newbie mistakes.
The problem I found though is that you can't really just use the raw filings. The data is much more useful when the stocks are queried with third party APIs and organized along with things like recent price data. This project alone uses three APIs, and while you could include that in a library and force the developer to get three different API keys, it just works better as a service.
If a "library" is all you want though, the API is available with documentation. There's also 13.info, an open-source project that predates this one. Although it is still a service, it is more like a library and could probably be used like one.