prompt
stringlengths 16
15.6k
| completion
stringlengths 4
6
|
|---|---|
Title: As the title implies, is there a site that tracks associations, lobbyists, etc of politicians?
Upvote: | 51 |
Title: Hi HN,<p>I built a little tool to help me get better at remembering country locations on the map: <a href="https://learn-worldmap.netlify.app/" rel="nofollow noreferrer">https://learn-worldmap.netlify.app/</a><p>It's made from plain HTML/CSS/JS, and not intended to make any money or anything. Just thought that maybe someone else may enjoy it. Also of course happy about any feedback!<p>Cheerz.
Upvote: | 45 |
Title: (I don't find them distracting and I enjoy messing with typography.)<p>I'm curious about any less obvious pros and cons, as well as recommendations for any fonts to try.
Upvote: | 42 |
Title: I thought it might be interesting to compile a list of things which it might be impossible (or at least very difficult) for a private individual to do under the Online Safety Bill.<p>Starters for ten:<p>1) Minecraft / MineTest server
2) IRC server
3) Mastodon server
4) BBS
5) ... please continue<p>I'm not fully sure about all this - as I understand it, if you allow chat or user generated content, you become a private service provider, and then have a host of responsibilities including annual audits of whether a significant proportion of your users are minors. If they are then you need to use a (commercial?) age verification tool and monitor everything assiduously. Difficult to see most people being able to satisfy those requirements.
Upvote: | 195 |
Title: Rain after long drought.<p><a href="https://www.digikey.com/en/products/detail/raspberry-pi/RASPBERRY-PI-4-MODEL-B-8G/12159401" rel="nofollow noreferrer">https://www.digikey.com/en/products/detail/raspberry-pi/RASP...</a> $75<p><a href="https://www.digikey.com/en/products/detail/raspberry-pi/SC1029/6152804" rel="nofollow noreferrer">https://www.digikey.com/en/products/detail/raspberry-pi/SC10...</a> $35
Upvote: | 51 |
Title: Hi HN! Over the last several weekends, I've been building LLMFlows as an alternative to langchain.<p>There's been a lot of discussion on the shortcomings of langchain in the past few weeks, but when I first tried it in March, I thought there are 3 main problems:
1. Too many abstractions
2. Hidden prompts and opinionated logic in chains which makes it hard to customize
3. Hard to debug<p>This inspired me to try and build a framework that solves these 3 issues, and therefore I started building LLFlows with the "philosophy" of being "simple, explicit, and transparent."<p>A few weekends later, I think I finally managed to reach a state where I feel it's ready to be shared.<p>I would love to hear your feedback! Thank you!
Upvote: | 49 |
Title: Hello HN,<p>What are some one man boring internet businesses you have encountered or know about
Upvote: | 76 |
Title: Suppose room temperature superconductor technology is validated and replicated. What to expect in the coming months, years and decades in the consumer space?
Upvote: | 51 |
Title: Hi folks, we're Debanjum and Saba. We created Khoj as a hobby project 2+ years ago because: (1) Search on the desktop sucked; we just had keyword search on the desktop vs google for the internet; and (2) Natural language search models had become good and easy to run on consumer hardware by this point.<p>Once we made Khoj search incremental, I completely stopped using the default incremental search (C-s) in Emacs. Since then Khoj has grown to support more content types, deeper integrations and chat (using ChatGPT). With Llama 2 released last week, chat models are finally good and easy enough to use on consumer hardware for the chat with docs scenario.<p>Khoj is a desktop application to search and chat with your personal notes, documents and images. It is accessible from within Emacs, Obsidian or your Web browser. It works with org-mode, markdown, pdf, jpeg files and notion, github repositories. It is open-source and can work without internet access (e.g on a plane).<p>Our chat feature allows you to extract answers and create content from your existing knowledge base. Example: <i>"What was that book Trillian mentioned at Zaphod's birthday last week"</i>. We personally use the chat feature regularly to find links, names and addresses (especially on mobile) and collate content across multiple, messy notes. It works online or offline: you can chat without internet using Llama 2 or with internet using GPT3.5+ depending on your requirements.<p>Our search feature lets you quickly find relevant notes, documents or images using natural language. It does not use the internet. Example: Search for <i>"bought flowers at grocery store"</i> will find notes about <i>"roses at wholefoods"</i>.<p>Quickstart:<p><pre><code> pip install khoj-assistant && khoj
</code></pre>
See <a href="https://docs.khoj.dev/#/setup">https://docs.khoj.dev/#/setup</a> for detailed instructions<p>We also have desktop apps (in beta) at <a href="https://github.com/khoj-ai/khoj/releases/tag/0.10.0">https://github.com/khoj-ai/khoj/releases/tag/0.10.0</a> if you want to try them out.<p>Please do try out Khoj and let us know if it works for your use cases? <i>Looking forward to the feedback!</i>
Upvote: | 565 |
Title: Hey folks! We're Alex and Evan, and we're working on putting together a 512 H100 compute cluster for startups and researchers to train large generative models on.
- it runs at the lowest possible margins (<$2.00/hr per H100)
- designed for bursty training runs, so you can take say 128 H100s for a week
- you don’t need to commit to multiple years of compute or pay for a year upfront<p>Big labs like OpenAI and Deepmind have big clusters that support this kind of bursty allocation for their researchers, but startups so far have had to get very small clusters on very long term contracts, wait months of lead time, and try to keep them busy all the time.<p>Our goal is to make it about 10-20x cheaper to do an AI startup than it is right now. Stable Diffusion only costs about $100k to train -- in theory every YC company could get up to that scale. It's just that no cloud provider in the world will give you $100k of compute for just a couple weeks, so startups have to raise 20x that much to buy a whole year of compute.<p>Once the cluster is online, we're going to be pretty much the only option for startups to do big training runs like that on.
Upvote: | 727 |
Title: I run a website whose domain is with GoDaddy. It is used by thousands of stores to take online orders. It was misused by a user for phishing. We banned him, but GoDaddy's compliance team has suspended the website on Friday evening and gone for the weekend. ETA should be less than 24 hours, but their support team is asking me to call on Monday morning. Hundreds of store owners are calling us, saying that they are not able to sell anything at the month-end. Do you know how to expedite the process?
Upvote: | 59 |
Title: TL;DR: <a href="https://pyflo.net" rel="nofollow noreferrer">https://pyflo.net</a> is a free, interactive guide to learning Python<p>Hi Everyone, I am a CS educator who has taught a variety of university courses, including many on introductory Python programming. Over the past few years, I've written down a number of Python programming lessons and has culminated into Pyflo.net. This tool is a completely free, introductory guide to learning Python. It is more-or-less an intro programming textbook, but with a few twists, including:<p>* It is totally free. You don't even have to give me your email to use it<p>* The lessons are short and modularized<p>* It's interactive, containing embedded questions that provide instant feedback throughout.<p>My hope is that this can be a useful resource for those looking to learn Python. Feel free to use yourself, or share with those you think would be interested.<p>Feedback is very much welcome and appreciated.
Upvote: | 181 |
Title: Hey folks! I'm Callum, and I'm working on a way to practice a new language with an AI powered tutor.<p>I've always found that the hardest part of learning a new language is finding someone to actually converse with. Even if a partner can be found, the pressure can mean that you are more focused on not making mistakes than on actually learning new grammar or vocabulary.<p>The service that I have been working on allows you to practice with a language tutor via online chat messages, or you can have a turn-based voice conversation.<p>I'm working on a number of other features that will be coming out shortly, including a few games for practising pronunciation and listening skills, as well as a plan to release some lesson plans for specific languages later on.<p>Have a try, and let me know if you have any feedback!
Upvote: | 157 |
Title: Hey HN. I wanted to share an update to our previous thread, “Notion with problem solving capabilities”.<p>The Decipad public beta is now live. You can try it for free here. <a href="https://www.decipad.com/" rel="nofollow noreferrer">https://www.decipad.com/</a><p>We started building Decipad to make numbers more expressive and playful. It’s a notebook environment where you can combine text, numbers, data and calculations into a story.<p>Our goal is to help people communicate with numbers more effectively and collaborate across diverse backgrounds. It’s feels a bit like Notion, but it’s for building interactive models and reports.<p>A few things we’ve been addressing building Decipad…<p>- A friendly modelling experience: You can express variables and calculations with quasi-natural language and connect them with tables, charts, pivot tables and other widgets.<p>- Unit expression: we built Decipad on a powerful unit system. You can assign labels and units to your data, like, `Cost = $5 per month per seat.`<p>- Dimensional Categories: Expressing relationships between variables and categories, making a model easy to adapt. We wrote about it here: <a href="https://www.decipad.com/blog/breaking-the-grid-overcoming-dimensional-constraints-in-spreadsheets" rel="nofollow noreferrer">https://www.decipad.com/blog/breaking-the-grid-overcoming-di...</a><p>- Connecting Data: Ability to connect data sources directly to your notebook. Right now, it’s intended for technical users. You can use JS and SQL to run a query.<p>We’re still exploring several areas like support for large data sets and building more UX interactions on top of our language to make modeling even more approachable and collaborative.<p>We would love to get feedback or any thoughts on our approach.
Upvote: | 88 |
Title: I have been the release manager for PyPy, an alternative Python interpreter with a JIT [0] since 2015, and have done a lot of work to make it available via conda-forge [1] or by direct download [2]. This includes not only packaging PyPy, but improving on an entire C-API emulation layer so that today we can run (albeit more slowly) almost the entire scientific python data stack. We get very limited feedback about real people using PyPy in production or research, which is frustrating. Just keeping up with the yearly CPython release cycle is significant work. Efforts to improve the underlying technology needs to be guided by user experience, but we hear too little to direct our very limited energy. If you are using PyPy, please let us know, either here or via any of the methods listed in [3].<p>[0] <a href="https://www.pypy.org/contact.html" rel="nofollow noreferrer">https://www.pypy.org/contact.html</a>
[1] <a href="https://www.pypy.org/posts/2022/11/pypy-and-conda-forge.html" rel="nofollow noreferrer">https://www.pypy.org/posts/2022/11/pypy-and-conda-forge.html</a>
[2] <a href="https://www.pypy.org/download.html" rel="nofollow noreferrer">https://www.pypy.org/download.html</a>
[3] <a href="https://www.pypy.org/contact.html" rel="nofollow noreferrer">https://www.pypy.org/contact.html</a>
Upvote: | 573 |
Title: I made a booking at Crowne Plaza Times Square on July 22, 2023. Set the option to show all prices and paid $1500 for it. At checking, in presented with a bill for city and state taxes and “resort fees” totaling $552 and a $500 security deposit. Doesn’t seem very transparent since they specially state that you’d see all taxes when you pay on Airbnb.<p>I called support when I was doing checkin, and they just said it was on the hotels terms/fine print in the listing and I should have read more carefully. Well, I assumed the value of Airbnb was to protect users from this, otherwise they’re just a middle man collecting a tax.<p>This is the listing btw. You’ll pay 30% more than advertised, so factor that in when comparing to other places. Also, weren’t security deposits unnecessary with Airbnb because they handle charging customers for things like damage, and aren’t listings insured?<p>Finally I use Airbnb a lot, so trust wise there should not be any red flags what’s justify security deposits.
Upvote: | 250 |
Title: Hey there HN!
Meet Linkwarden, a fully self-hostable, open-source collaborative bookmark manager to collect, organize and archive webpages.<p>Please also visit/star our GitHub repo [1].<p>Linkwarden was built using TypeScript and NextJS, backed by a PostgreSQL database for the lighter-weight data. The rest of the data can be chosen either to be stored on the filesystem, or stored on the cloud on Digital Ocean Space/AWS S3, the reason for the cloud storage solution was for the Cloud offering [2], we realized that the preserved webpages (archives) take up space pretty quickly and S3 was much more efficient for this task. On the front-end we used TailwindCSS for styling and Zustand for state management.<p>You could either use our Cloud offering (with 14-day free trial) to directly support this project and experience Linkwarden, or you could self-host it on your own machine and have maximum flexibility.<p>Feel free if you had any questions, we'll do our best to answer it.<p>[1]: <a href="https://github.com/linkwarden/linkwarden">https://github.com/linkwarden/linkwarden</a><p>[2]: <a href="https://cloud.linkwarden.app/register" rel="nofollow noreferrer">https://cloud.linkwarden.app/register</a> - Hosted in Digital Ocean's datacenter located here in Toronto, ON.
Upvote: | 280 |
Title: I've been working on markwhen for a bit as a way to create timelines and calendars from plain text, like markdown.<p>I personally like tools that let you immediately start using them, and I set out to do that here with markwhen.<p>Let me know if you have any questions or feedback!
Upvote: | 327 |
Title: I am using systemd on my machine and try to configure most things through it. For example, I have a backup job that is triggered by a timer.
I want to know when that job fails so I can investigate and fix it. Over time, I've had multiple solutions for this:<p>Send a notifcation via notify-send<p>Add `systemctl --failed` to my shell startup script<p>Send myself emails<p>None of these are quite ideal. Notifications are disruptive of the current workflow and ephemeral, meaning I might forget about it if I don't deal with it immediately.
Similarly, reading `systemctl --failed` on every new terminal is also disruptive but at least it makes me not forget about it.
Both of these are also not really applicable to server systems.
Sending myself emails feels a bit wrong but has so far been the best solution.<p>How are other people solving this? I did some research and I am surprised that there isn't a more rounded solution. I'd expect that pretty much every Linux user must run into this problem.
Upvote: | 127 |
Title: Warning: NSFW.<p>I'm posting this from a throwaway. Note that on HN favorites are public, but upvotes are private.<p>Porn seems like it'll be one of the big areas of ai in the short term. There've been multiple threads and comments about AI + Onlyfans and AI + porn and AI girlfriends on HN recently (<a href="https://news.ycombinator.com/item?id=36888038">https://news.ycombinator.com/item?id=36888038</a>, <a href="https://news.ycombinator.com/item?id=36920987">https://news.ycombinator.com/item?id=36920987</a>, <a href="https://news.ycombinator.com/item?id=36849066">https://news.ycombinator.com/item?id=36849066</a>).<p>I've heard that porn is the industry that pushes new technology forward - VHS, bluray, highly scalable video serving.<p>I made a site compiling together some of the AI porn that I made with the help of some people on Discord over the past few months.<p>I don't think AI will get rid of onlyfans. People spend on onlyfans for more than just the images.<p>I do think people will have access to unlimited image and video porn at some point in the coming decade, which will definitely change the porn industry.<p>The site shows what's possible with AI porn today - images both sfw and nsfw, audio, text, and some moving images. There's even more possible than just what's on the page, with lots of people are playing around with making animated gifs that are nsfw, there'll be animated 3d nsfw avatars, and so on. So the site isn't comprehensive at all. But it's still quite interesting.
Upvote: | 113 |
Title: This one is wild. I ordered some medication from Amazon as I usually do. The medication gets dropped off this afternoon around 3PM. At around 7PM the driver from this morning (Intelcom) sends me a text, asking to give him a 5 star review. Amazon is giving complete customer profiles to third party courier companies and is denying it on the phone and in writing.<p>"Hi this is your Intelcom driverYour package was delivered today by me. If you liked my service please give me a 5 star rating in your emails from "Intelcom". It will mean a lot to me! Have a good one :)"<p>Finally, I've connected with a "supervisor", who is saying that drivers texting from personal lines after hours is totally normal, and that they're not permitted to use their cell phone while driving as it's a safety concern.<p>Anyway, I'm now more upset about it than I should be because I've wasted 2 hours already with Amazon on this.
Upvote: | 59 |
Title: Hi HN! I'm Jarek, and I've built this tool that allows publishing .local domains on the local network using mDNS.<p>It also has a reverse proxy that handles HTTPS termination and port forwarding.<p>I'm working on adding more features, like an index page with all available domains or allowing proxy redirects, so you could redirect from HTTP to HTTPS.<p>Let me know if you have any questions or feedback!
Upvote: | 84 |
Title: Has anyone else encountered this phenomenon lately? I've found myself prompting GPT 3.5 with simple questions that GPT 4 provided an incorrect answer for, and lo and behold I get a much better answer.<p>For ex this is GPT 4: https://chat.openai.com/share/e24501ad-8f1c-4b5a-a6d0-d933f5d1d209<p>And this is GPT 3.5: https://chat.openai.com/share/b9372bdc-ffff-4655-bee4-2b3f3c3b8285<p>In the latter case I didn't even need to ask for the order by clause as it anticipates it and provides an answer for it. GPT 4's first answer was wrong.<p>In the past two days I've seen at least 2 other cases where GPT 4's answer was plain wrong and GPT 3.5's was not only correct but of very high quality, reminding me of what I first felt when using GPT 4 for the first time.
Upvote: | 51 |
Title: Please state the location and include REMOTE, INTERNS and/or VISA
when that sort of candidate is welcome. When remote work is <i>not</i> an option,
include ONSITE.<p>Please only post if you personally are part of the hiring company—no
recruiting firms or job boards. One post per company. If it isn't a household name,
explain what your company does.<p>Commenters: please don't reply to job posts to complain about
something. It's off topic here.<p>Readers: please only email if you are personally interested in the job.<p>Searchers: try <a href="https://hnjobs.u-turn.dev" rel="nofollow noreferrer">https://hnjobs.u-turn.dev</a>, <a href="https://hnresumetojobs.com" rel="nofollow noreferrer">https://hnresumetojobs.com</a>, <a href="https://hnhired.fly.dev" rel="nofollow noreferrer">https://hnhired.fly.dev</a>,
<a href="https://kennytilton.github.io/whoishiring/" rel="nofollow noreferrer">https://kennytilton.github.io/whoishiring/</a>, <a href="https://hnjobs.emilburzo.com" rel="nofollow noreferrer">https://hnjobs.emilburzo.com</a>.<p>Don't miss these other fine threads:<p><i>Who wants to be hired?</i> <a href="https://news.ycombinator.com/item?id=36956865">https://news.ycombinator.com/item?id=36956865</a><p><i>Freelancer? Seeking freelancer?</i> <a href="https://news.ycombinator.com/item?id=36956866">https://news.ycombinator.com/item?id=36956866</a>
Upvote: | 479 |
Title: Please lead with either SEEKING WORK or SEEKING FREELANCER,
your location, and whether remote work is a possibility.
Upvote: | 57 |
Title: Share your information if you are looking for work. Please use this format:<p><pre><code> Location:
Remote:
Willing to relocate:
Technologies:
Résumé/CV:
Email:
</code></pre>
Readers: please only email these addresses to discuss work opportunities.
Upvote: | 201 |
Title: Hi HN! We're Nick, Patrick, Philip, Sebastian, Sven, and Timo from Titap (<a href="https://tiptap.dev/">https://tiptap.dev/</a>), an open source developer toolkit for building collaborative editing apps. Our editor framework, based on ProseMirror, is at <a href="https://github.com/ueberdosis/tiptap">https://github.com/ueberdosis/tiptap</a>, and our real-time collaboration backend, based on Yjs, is at <a href="https://github.com/ueberdosis/hocuspocus">https://github.com/ueberdosis/hocuspocus</a>.<p>Building editor interfaces like Notion or Google Docs in your web app takes a lot of work and time. Our open source tools and cloud services let you build collaborative content editing faster—in days or weeks, rather than months or years. And this is just for the editor. If you want real-time collaboration or other advanced features like version history in your editor, the overall workload quickly escalates—you will need a robust and serious backend infrastructure that requires even more time to set up and maintain. This doesn’t make sense for most frontend developers or most startups.<p>We spent eight years as a digital agency developing applications with complex content editing functionality. We learned the hard way how limited the existing editors were. After building Tiptap as a headless editor framework with an extension-based architecture, we needed to allow multiple users to edit content simultaneously, which got complicated. There was no simple solution that could be integrated quickly. So we built that too.<p>The Tiptap editor is based on the JS framework ProseMirror, which is a good foundation for editors. The learning curve for ProseMirror is steep because it's complicated to understand and lacks simple APIs and documentation. It takes a lot of code around ProseMirror to develop a modern user experience. We’ve taken care of that for you.<p>Tiptap is headless, so it will work with whatever frontend or design you have in mind—we make no assumptions about your UI. You can use it to develop block-based editors like Notion, classic interfaces like Google Docs, or whatever you need. It's also framework agnostic, so you can use it with React, Vue, etc., or vanilla JavaScript. And it's highly customizable through our extension architecture. We also provide an API to access ProseMirror's internals through Tiptap if you want to dig deep into the core.<p>Adding real-time collaboration to your editor is as easy as installing and configuring an extension. Our collaboration backend, called Hocuspocus, uses Yjs. This is a widely used implementation of CRDTs (conflict- free replicated data type). Hocuspocus makes it easy to set up a Node.js websocket server to handle communication between multiple peers to synchronize data. Like the Tiptap editor, Hocuspocus is designed to be extensible according to your needs. Also, Hocuspocus can work independently of Tiptap with other editors like Lexical or Slate.<p>An earlier version of Tiptap got discussed a couple years ago at <a href="https://news.ycombinator.com/item?id=26901975">https://news.ycombinator.com/item?id=26901975</a>. We’ve been enjoying wider adoption since then. For example, Substack uses Tiptap for their editor that allows creators to write content on substack.com, and YC uses Tiptap in their Bookface forum (which is basically HN for YC alums).<p>With the Tiptap Cloud, we offer managed backend services if you don't want to build and maintain every feature yourself. For real-time collaboration, we provide a cloud infrastructure with multiple datacenter regions where you can deploy Hocuspocus. The Tiptap AI integration beta is a service where you connect your OpenAI API key to our backend and install the Tiptap editor AI extension to get AI writing experience in your editor. Here’s a demo: <a href="https://ai-demo.tiptap.dev/">https://ai-demo.tiptap.dev/</a><p>We invite you to explore Tiptap's capabilities in your app, contribute to its open source development, and (hopefully!) join our welcoming community. We'd love to hear what you've already built with Tiptap or what's stopping you from creating something with it :-) We look forward to all of your comments!
Upvote: | 246 |
Title: Hey HN! We’re Kevin and Steve. We’re building PromptTools (<a href="https://github.com/hegelai/prompttools">https://github.com/hegelai/prompttools</a>): open-source, self-hostable tools for experimenting with, testing, and evaluating LLMs, vector databases, and prompts.<p>Evaluating prompts, LLMs, and vector databases is a painful, time-consuming but necessary part of the product engineering process. Our tools allow engineers to do this in a lot less time.<p>By “evaluating” we mean checking the quality of a model's response for a given use case, which is a combination of testing and benchmarking. As examples:
- For generated JSON, SQL, or Python, you can check that the output is actually JSON, SQL, or executable Python.
- For generated emails, you can use another model to assess the quality of the generated email given some requirements, like whether or not the email is written professionally.
- For a question-answering chatbot, you can check that the actual answer is semantically similar to an expected answer.<p>At Google, Steve worked with HuggingFace and Lightning to support running the newest open-source models on TPUs. He realized that while the open-source community was contributing incredibly powerful models, it wasn’t so easy to discover and evaluate them. It wasn’t clear when you could use Llama or Falcon instead of GPT-4. We began looking for ways to simplify and scale this evaluation process.<p>With PromptTools, you can write a short Python script (as short as 5 lines) to run such checks across models, parameters, and prompts, and pass the results into an evaluation function to get scores. All these can be executed on your local machine without sending data to third-parties. Then we help you turn those experiments into unit tests and CI/CD that track your model’s performance over time.<p>Today we support all of the major model providers like OpenAI, Anthropic, Google, HuggingFace, and even LlamaCpp, and vector databases like ChromaDB and Weaviate. You can evaluate responses via semantic similarity, auto-evaluation by a language model, or structured output validations like JSON and Python. We even have a notebook UI for recording manual feedback.<p>Quickstart:<p><pre><code> pip install prompttools
git clone https://github.com/hegelai/prompttools.git
cd prompttools && jupyter notebook examples/notebooks/OpenAIChatExperiment.ipynb
</code></pre>
For detailed instructions, see our documentation at <a href="https://prompttools.readthedocs.io/en/latest/" rel="nofollow noreferrer">https://prompttools.readthedocs.io/en/latest/</a>.<p>We also have a playground UI, built in streamlit, which is currently in beta: <a href="https://github.com/hegelai/prompttools/tree/main/prompttools/playground">https://github.com/hegelai/prompttools/tree/main/prompttools...</a>. Launch it with:<p><pre><code> pip install prompttools
git clone https://github.com/hegelai/prompttools.git
cd prompttools && streamlit run prompttools/ui/playground.py
</code></pre>
We’d love it if you tried our product out and let us know what you think! We just got started a month ago and we’re eager to get feedback and keep building.
Upvote: | 211 |
Title: For example:<p>I've used 3+ code editors on MacOS and prefer sublime text over VScode, coteditor, xcode.<p>I've used Chrome, Firefox and Safari all within the past year, and I prefer Chrome and Safari.<p>I've used 3+ voice transcription apps, the ones I prefer are 'just press record' and Otter.ai, I can't remember the names of the others I used - but I downloaded a bunch of Whisper based ones and non-whisper based ones on the iOS app store.<p>I've used 3+ messaging apps, I like iMessage and Telegram, I prefer those over Signal and WhatsApp.<p>I've used 3+ interfaces for GPT, and I prefer the OpenAI playground over ChatGPT, chatbot-ui, typingmind, openplayground, Poe, and a few others I can't remember.<p>I think it's important to note that for me these are all preferences. I'm not saying one of these tools are objectively better. I am saying that for me they are better, and I prefer them.<p>What are areas of products, tools, developer tools, APIs, anything, where you've personally used 3+ tools, and what did you prefer, and what did you not prefer, if you can still remember?
Upvote: | 127 |
Title: Howdy! We built this as an experiment in personal-programming, combining the best of LLMs and code to help automate tasks around you. I personally use it to track the tides and get notified when certain conditions are met, something that pure LLMs had trouble dealing with and pure code was often too brittle for.<p>We created it after getting frustrated with the inability of LLMs to deal with numbers and the various hoops we had to jump through to make ChatGPT output repeatable.<p>At the core, Magic Loops are just a series of "blocks" (JSON) that can be triggered with different inputs (email, time, webhook), then operate on those inputs using a combination of LLMs and code, and then output those results (email, text, webhook). Under the hood, the LLM calls are using GPT-4 via OpenAI and the code is run in sandboxed (no internet) Docker containers in AWS.<p>You have full control over each step of the loop, but you can also create (or attempt to create) a Magic Loop by simply describing what you want. We use GPT-4 to break that request into feasible steps, and then create a Magic Loop scaffold. Of course, you should still validate the loop before publishing it!<p>We've seen some neat use cases already:<p>- "Text me when the tide is less than 1ft between 7am and 7pm at Fort Funston"<p>- "Summarize an email using this format and forward it to this address"<p>- "Text me every time our store does more than $1000/day in volume on Shopify"<p>- "Take specific data from Cloudflare, format it, and send it to Mixpanel every hour"<p>We hope you enjoy what's essentially an experiment at this point. If folks like the concept, we're thinking about open sourcing it so you can run the loops locally with the code runtimes you wish (rather than in our code runners).<p>Let us know what you think, and more importantly, what you wish to build or automate!<p>Cheers,
Adam & Mihai
Upvote: | 248 |
Title: Yes, I said it, Stripe is no longer a suitable payment processor unless you're a medium to huge sized company to which you have some weight in what you say. small business owners who clearly have good standing to which stripe has held funds, blocked and shutdown accounts (including my own) are all complaining.. there is a clear pattern here -> https://uk.trustpilot.com/review/stripe.co.uk?stars=1
https://www.sitejabber.com/reviews/stripe.com<p>They might as well only accept huge companies only because they're surviving off their top 10, their support is terrible, the fraud detection is terrible and when you make a decent amount of sales they will shut down your account. I urge you to try it yourself on a newly made account, guarantee you run into some problems, stripe is 2023 is a joke
Upvote: | 202 |
Title: It is not clear whether relatively small open source LLMs in production or not, for instance replit code 3b, llama 2 7b, codegen,... What are the motivations for using those models over prompted GPT api ?
Upvote: | 50 |
Title: Hi HN! We’re Kevin and Daryl from Certainly Health (<a href="https://certainlyhealth.com/">https://certainlyhealth.com/</a>), a marketplace that lets you shop for doctors and compare your out-of-pocket costs upfront (currently only in NYC). Using each patient's insurance and machine learning, we predict and guarantee out-of-pocket costs to prevent surprise bills.<p>We grew up in immigrant families, and for both of us, our parents would honestly tell us to not trust doctors because of unexpected medical bills. They really believed that doctors were adding on unnecessary things that didn’t help in order to make more money. As adults, we’ve both had first-hand experiences receiving surprise bills even with good insurance while working at tech companies, so solving this problem is personal to us.<p>It turns out 38% of Americans delay medical care for fear of the bill while prices vary 2-10x across health providers, even when using insurance. This led us to believe that creating a marketplace with transparent out-of-pocket costs could be a solution.<p>However, it wasn’t until the July 2022 Transparency in Coverage Rule (<a href="https://www.cms.gov/healthplan-price-transparency" rel="nofollow noreferrer">https://www.cms.gov/healthplan-price-transparency</a>) that we had the payer pricing data (that’s actually high quality compared to hospital data) to create this marketplace. Certainly is the first company to use this data to predict and guarantee prices for consumers to book doctors.<p>We know the HN community has been interested in applications of healthcare price transparency data (<a href="https://news.ycombinator.com/item?id=32738783">https://news.ycombinator.com/item?id=32738783</a>), so we’re excited to share how we’re using this data to predict and guarantee prices for a consumer application.<p>With Certainly, patients can enter their insurance, compare guaranteed out-of-pocket costs across doctors, and book an appointment with price transparency. For providers, we guarantee any patient we send them will always pay. Partner providers pay us any time we send them a new patient. Eventually, we’ll convert providers to using our SaaS platform to guarantee payments from existing patients.<p>We are currently focused on shoppable services - anything that can be scheduled in advance.
Our solution falls between two ends of the spectrum for consumer shopping experiences. At one end is healthcare, where the status quo is you go to a doctor and you have no idea what will happen and you have no idea how much it will cost. At the other end is going to a restaurant, having a menu of items with corresponding prices, and deciding which items you want to order.<p>Certainly Health lets you see what services could happen during your visit and how much they will cost. We guarantee that if those services happen, you will not pay more than the out-of-pocket costs you saw upfront for each of those services. But it is ultimately the doctor who decides which services you end up getting (like a chef deciding which dishes you will be served). As a result, you can compare upfront out-of-pocket costs specific to your insurance and book a provider without worrying about surprise bills.<p>A common misconception is that patient out-of-pocket costs are set by doctors. Prices of healthcare services are actually the result of negotiations between providers and insurance companies. Groups with more negotiating power, like large hospital systems, are able to command higher rates than private practice physicians. This variance in cost is enormous across almost all procedures we've looked at, even between providers a few blocks from each other. It might cost $105 to get an orthopedic consultation with one doctor, and $550 to get a consultation with another doctor across the street, for example. This price variance means the out-of-pocket cost also varies for patients with high deductible plans or plans where the copay does not cover all services (which we’ve encountered very often).<p>We do not use hospital price data which is inconsistent and messy (<a href="https://www.kff.org/health-costs/press-release/analysis-inconsistencies-within-hospital-price-transparency-data-make-cost-comparisons-difficult/" rel="nofollow noreferrer">https://www.kff.org/health-costs/press-release/analysis-inco...</a>). Instead, our guaranteed prices are based on three things: (1) health insurance published rates for procedures, as required by regulation; (2) patient eligibility/benefits information (e.g. deductible, copay, coinsurance); and (3) an online learning model to account for variance across thousands of plans and patient conditions.<p>As we process payments and claims, we create an accurate platform for predicting out-of-pocket costs by accounting for variations such as whether certain procedures are covered by an insurance plan/copay, and how the rate changes based on diagnosis code (we discovered V97.33XD is for "Sucked into jet engine, subsequent encounter”), number of units, medical coding modifiers, and other factors.<p>Certainly is currently available in NYC where you can choose from over 5k doctors to book without surprise bills for free at <a href="https://certainlyhealth.com/">https://certainlyhealth.com/</a>. We plan to expand to other cities once we prove that we can establish a low CAC on both the supply and demand sides in NYC.<p>Although we’re early, we’re starting to see intended results. One patient saved $850 on a podiatrist visit booked through us compared to a previous visit with a different doctor for the exact same services. Another was charged for an annual physical, but we identified the mistake and got a refund issued to him. Most of our customers have had surprise bills in the past and are happy to have peace of mind with our upfront, guaranteed prices.<p>We’d love to hear your ideas, experiences, feedback and any feature requests!
Upvote: | 186 |
Title: Hi to all.<p>My name is Luca and I’m writing this thread on behalf of my business for a problem we are currently facing with our Stripe account.<p>To sum up our story:
- We created our business Stripe account months back…no complaints.
- 2 weeks ago, we opened the payment gates and received the first 10 payments after 1 year of work…account disabled for no reason and with no explanation.
- We went through multiple support tickets and emails with Stripe support and got the same answer every time from, what seems to be, a chatbot.
- We contacted the "Specialist team" through LinkedIn and after thoroughly explaining how we do not fall under any Restricted Business written by Stripe, he still decided to close our account for no reason and with no explanation.
- The Stripe team is ignoring the fact that we fall under the same category as our competitor, which literally offers our same exact service. In the end, they still decided to close our account...again, for no reason and with no explanation.<p>Now after 1 year of work and weeks of integrating Stripe services to our platform, clients’ money involved, what are we supposed to do?
Does Stripe at least have a support to pass clients recurring payments to another platform?
Or do we just lose all of our clients?<p>If there’s anything my team can do such as: changing the service, the platform or the landing page to become compliant to Stripe’s Policies please help us out.<p>If anyone knows someone we can hop on a call with or discuss this situation please leave details or contact me at:<p>[email protected]
Upvote: | 46 |
Title: Hi guys,<p>Hope everyone is well. This app was borne out of my own frustration. I thought that I was terrible at learning languages at school, since I didn't become conversational in French after 5 years of study. However, I later traveled with some French friends and, in just under 3 weeks, I was able to hold a reasonable conversation. I realized that there's no substitute for speaking to native speakers.<p>I tried to adopt this approach for other languages, but it's much harder to find people to practise with when you aren't travelling. I started using iTalki to meet people from different countries and chat to them. It quickly became very expensive and time-consuming to schedule the calls, so I gave up.<p>I made PrettyPolly so that anyone can easily practice speaking 26 languages orally. The app uses ChatGPT (amongst other tools) to allow you to practice speaking whenever you want. It also generates a fluency score for each conversation so that you have an objective way of tracking progress.<p>It's free to use (up to 15 conversations per month). I've found that using it once or twice per day is plenty, and you'll be amazed at how much you will pick up in a week. I've added some FAQs here in case useful - <a href="https://www.prettypolly.app/learn" rel="nofollow noreferrer">https://www.prettypolly.app/learn</a><p>Would really appreciate any feedback. Let me know if you have any questions, issues or suggestions.<p>Thanks,
Chris
Upvote: | 325 |
Title: After more than 9 years of using them for my SaaS business, Stripe has decided to close my account for the following reason:<p>"""Unfortunately, after conducting a further review of your account, we’ve confirmed that your business presents a higher level of risk than we can currently support, so we won’t be able to accept payments for XYZ LLC moving forward."""<p>This is after I've made drastic changes that help reduce chargebacks and asked for a review.<p>I've faded all posts and complaints here against Stripe thinking that people <i>must</i> be up to no good. And that it can't happen to me...
Upvote: | 61 |
Title: If so, which one?
Upvote: | 91 |
Title: I've been disappointed by the very poor quality of results that I generally get when trying to run OCR on older scanned documents, especially ones that are typewritten or otherwise have unusual or irregular typography. I recently had the idea of using Llama2 to use common sense reasoning and subject level expertise to correct transcription errors in a "smart" way-- basically doing what a human proofreader who is familiar with the topic might do.<p>I came up with the linked script that takes a PDF as input, runs Tesseract on it to get an initial text extraction, and then feeds this sentence-by-sentence to Llama2, first to correct mistakes, and then again on the corrected text to format it as markdown where possible.
This was surprisingly easier than I initially expected thanks to the very nice tooling now available in libraries such as llama-cpp-python, langchain, and pytesseract. But the big issue I was encountering was that Llama2 wasn't just correcting the text it was given-- it was also hallucinating a LOT of totally new sentences that didn't appear in the original text at all (some of these new sentences used words which never appeared elsewhere in the original text).<p>I figured this would be pretty simple to filter out using fuzzy string matching-- basically check all the sentences in the LLM corrected text and filter out sentences that are very different from any sentences in the original OCRed text. To my surprise, this approach worked very poorly. In fact, lots of other similar tweaks, including using bag-of-words and the spacy NLP library in various ways (spacy worked very poorly in everything I tried).<p>Finally I realized that I had a good solution staring me in the face: Llama2. I realized I could get sentence level vector embeddings straight from Llama2 using langchain. So I did that, getting embeddings for each sentence in the raw OCRed text and the LLM corrected text, and then computed the cosine similarity of each sentence in the LLM corrected text against all sentences in the raw OCRed text. If no sentences match in the raw OCRed text, then that sentence has a good chance of being hallucinated.<p>In order to save the user from having to experiment with various thresholds, I saved the computed embeddings to an SQLite database so they only had to be computed once, and then tried several thresholds, comparing the length of the filtered LLM corrected text to the raw OCRed text; if things worked right, these texts should be roughly the same length. So as soon as the filtered length dips below the raw OCRed text length, it backtracks and uses the previous threshold as the final selected threshold.<p>Anyway, if you have some very old scanned documents laying around, you might try them out and see how well it works for you. Do note that it's extremely slow, but you can leave it overnight and maybe the next day you'll have your finished text, which is better than nothing! I feel like this could be useful for sites like the Internet Archive-- I've found their OCR results to be extremely poor for older documents.<p>I'm very open to any ideas or suggestions you might have. I threw this together in a couple days and know that it can certainly be improved in various ways. One idea that I thought might be fun would be to make this work with a Ray cluster, sending a different page of the document to each of the workers in the cluster to do it all at the same time.
Upvote: | 54 |
Title: Recently there was a thread about a "breakthrough" in battery technology at Toyota.<p><a href="https://news.ycombinator.com/item?id=36585327">https://news.ycombinator.com/item?id=36585327</a><p>Toyota has been putting out PR puff pieces about their "solid-state" (solid-electrolyte) batteries for years, but this story was unique in that it had a quote from Keiji Kaita, who holds some high-level role at Toyota. Anyway, I didn't think much of it, because there was no paper referenced in the <i>Guardian</i> article, which seemed to be the original source.<p>But while reading about something else, I came across the paper "A near dimensionally invariable high-capacity positive electrode material", published in <i>Nature Materials</i> last December:<p><a href="https://www.nature.com/articles/s41563-022-01421-z" rel="nofollow noreferrer">https://www.nature.com/articles/s41563-022-01421-z</a><p>This paper, reporting a cathode that has very little (much less than normal) change in size or shape when charged and discharged, claims reversible storage with a solid electrolyte. It stands to reason that dimensional stability of the cathode is necessary for interfacing with a solid electrolyte, since if it swells and shrinks, it will probably detach from the electrolyte, and possibly damage it further.<p>Looking at the affiliations of some of the authors we see a number of contributors from the "Lithium Ion Battery Technology and Evaluation Center (LIBTEC)". A web search about LIBTEC leads to several articles from 2018:<p><a href="https://www.cnet.com/roadshow/news/toyota-nissan-honda-libtec-solid-state-battery-development/" rel="nofollow noreferrer">https://www.cnet.com/roadshow/news/toyota-nissan-honda-libte...</a><p>which state that Toyota, along with Nissan, Honda and Panasonic (Tesla's major collaborator), have established this consortium to work on solid-electrolyte batteries as of five years ago.<p>So what does this thing look like? It's a vanadium–titanium cathode, Li8Ti2V4O14. Titanium is common; vanadium technically has a higher crustal abundance than nickel, but it tends to be spread across low-quality deposits, so production is low right now. A review considering the resource outlook for V-based batteries [1] was guardedly optimistic. 750 Wh/kg is <i>great</i>. Vanadium cathodes historically had a problem with high dimensional <i>instability</i>, but it appears that cocrystallization with titanium may have fixed that, and the weird properties of vanadium became an advantage in compensating for Li+ influx/efflux.<p>The use of a sulfide electrolyte pours doubt on claims of safety, though. It's reasonably likely that if water were to come into contact with the electrolyte, it could release highly toxic hydrogen sulfide gas.<p>Also, since the battery was developed in collaboration with other major automakers (and funded by the Japanese government), it's somewhat questionable to think it would give Toyota a major advantage in the EV race. But for the Japanese economy, which has been rather slow lately, it could be a boost.<p>1: <a href="https://chemistry-europe.onlinelibrary.wiley.com/doi/abs/10.1002/cssc.202200479" rel="nofollow noreferrer">https://chemistry-europe.onlinelibrary.wiley.com/doi/abs/10....</a>
Upvote: | 529 |
Title: Well, the title says it all..
Upvote: | 91 |
Title: Hi HN! We’re William and Kevin, cofounders of Sweep (<a href="https://sweep.dev/">https://sweep.dev/</a>). Sweep is an open-source AI-powered junior developer. You describe a feature or bugfix in a GitHub issue and Sweep writes a pull request with code. You can see some examples here: <a href="https://docs.sweep.dev/examples">https://docs.sweep.dev/examples</a>.<p>Kevin and I met while working at Roblox. We talked to our friends who were junior developers and noticed a lot of them doing grunt work. We wanted to let them focus on important work. Copilot is great, but we realized some tasks could be completely offloaded to an AI (e.g. adding a banner to your webpage <a href="https://github.com/sweepai/landing-page/issues/225">https://github.com/sweepai/landing-page/issues/225</a>).<p>Sweep does this with a code search engine. We use code chunking, ranking, and formatting tricks to represent your codebase in a token-efficient manner for LLMs. You might have seen our blog on code chunking here: <a href="https://news.ycombinator.com/item?id=36948403">https://news.ycombinator.com/item?id=36948403</a>.<p>We take these fetched code snippets and come up with a plan to write the PR. We found that having the LLM provide structured information using XML tags is very robust, as it’s easy for us to parse with regex, has good support for multi-line answers and is hard for the LLM to mess up.<p>This is because XML is common in the LLM’s training data (the internet / HTML), and the opening and closing tags rarely appear naturally in text and code, unlike the quotations, brackets, backticks and newlines used by JSON’s and markdown’s delimiters. Further, XML lets you skip the preamble (“This question has to do with xyz. Here is my answer:”) and handles multi-line answers like PR plans and code really well. For example, we ask the LLM for the new code in <new_code> tags and a final boolean answer by writing <answer>True</answer>.<p>We use this XML format to get the LLM to create a plan, generating a list of files to create and modify from the retrieved relevant files. We iterate through the file changes and edit/create the necessary files. Finally, we push the commits to GitHub and create the PR.<p>We’ve been using Sweep to handle small issues in Sweep’s own repo (it recently passed 100 commits). We’ve become well acquainted with its limitations. For example, Sweep sometimes leave unimplemented functions with just “# rest of code” since it runs on GPT-4, a model tuned for chatting. Other times, there’s minor syntax errors or undefined variables. This is why we spend the other half of our time building self-recovery methods for Sweep to fix and test its PRs.<p>First, we invite the developer to review and add comments to Sweep’s pull request. This helps to a point, but Sweep’s code sometimes wouldn’t lint. This is table stakes. It’s frustrating to have to tell the bot to “add an import here” or “this variable is undefined”.
To make this better, we used GitHub Actions, which automatically runs the flow of “check the code → tell sweep → sweep fixes the code → check the code again”. We like this flow because you might already have GitHub Actions, and it’s fully configurable. Check out this blog to learn more <a href="https://docs.sweep.dev/blogs/giving-dev-tools">https://docs.sweep.dev/blogs/giving-dev-tools</a>.<p>So far, Sweep isn’t that fast, can’t handle massive problems yet, and doesn’t write hundreds of lines of code. We’re excited to work towards that. In the meantime, a lot of our users have been able to get useful results. For example, a user reported that an app was not working correctly on Windows, and Sweep wrote the PR at <a href="https://github.com/sweepai/sweep/pull/368/files">https://github.com/sweepai/sweep/pull/368/files</a>, replacing all occurrences of "/tmp" with "tempfile.gettempdir()". Other examples include adding a validation function for Github branch name (<a href="https://github.com/sweepai/sweep/pull/461">https://github.com/sweepai/sweep/pull/461</a>) and adding dynamically generated initials in the testimonials on our landing page (<a href="https://github.com/wwzeng1/landing-page/issues/28">https://github.com/wwzeng1/landing-page/issues/28</a>). For more examples, checkout <a href="https://docs.sweep.dev/examples">https://docs.sweep.dev/examples</a>.<p>Our focus is on finding ways that an AI dev can actually help and not just be a novelty. I think of my daily capacity to write good code as a stamina bar. There’s a fixed cost to opening an IDE, finding the right lines of code, and making changes. If you’re working on a big feature and have to context switch, the cost is higher. I’ve been leaving the small changes to Sweep, and my stamina bar stays full for longer.<p>Our repo is at <a href="https://github.com/sweepai/sweep">https://github.com/sweepai/sweep</a>, there’s a demo video at <a href="https://www.youtube.com/watch?v=WBVna_ow8vo">https://www.youtube.com/watch?v=WBVna_ow8vo</a>, and you can install Sweep here: <a href="https://github.com/apps/sweep-ai">https://github.com/apps/sweep-ai</a>. We currently have a freemium model, with 5 GPT-4 PRs at the free tier, 120 GPT-4 PRs at the paid tier and unlimited at the enterprise tier.<p>We’re far from our vision of a full AI software engineer, but we’re excited to work on it with the community feedback :). Looking forward to hearing any of your thoughts!
Upvote: | 198 |
Title: hi hn, hydra ceo here<p>hydra is an open source, column-oriented postgres. you can set up remarkably fast aggregates on your project in minutes to query billions of rows instantly.<p>postgres is great, but aggregates can take minutes to hours to return results on large data sets. long-running analytical queries hog database resources and degrade performance. use hydra to run much faster analytics on postgres without making code changes. data is automatically loaded into columnar format and compressed. connect to hydra with your preferred postgres client (psql, dbeaver, etc).<p>following 4 months of development on hydra v0.3.0-alpha, our team is proud to share our first major version release. hydra 1.0 is under active development, but ready for use and feedback. we’re aiming to release 1.0 into general availability (ga) soon.<p>for testing, try the hydra free tier to create a column-oriented postgres instance on the cloud. <a href="https://dashboard.hydra.so/signup">https://dashboard.hydra.so/signup</a>
Upvote: | 269 |
Title: I have about 3 Yoe webdev/+physics/mech eng degree. Worked a few contracts then my own Saas thing, now pretty burned out. Don't really need funding or a team, just a bit of time.<p>Was thinking to just to dumb manual labour while I recover/finish my product. I don't think I can hack a 3 month job hunt for a "legit" job and I feel like taking tech contracts will also create a lot of mental overhead since I'm in the headspace for my own work right now.<p>Maybe tutoring would be better than Ubereats gig work, but honestly a lot of software contracts seem to be turned into BS gig work and at least with manual labour you are doing something real.<p>Looking for validation and ideas.
Upvote: | 62 |
Title: I'm Twitter user number 80 thousand something. I have a two-letter twitter username @tv. Have had it for ages. It's my initials, and a nickname I've been called for decades. I just got this in email:<p>--8<--<p>Hello,<p>The user handle associated with account @TV will be affiliated with X Corp. Accordingly, your user handle will be changed to a new user handle.<p>However, we appreciate your loyalty and want to minimize any inconvenience this may cause. At this time we will be changing this handle to @TheTV. We have listed additional handles that you can choose from below. Just respond to this message and we can assist you in making the change. All data associated with your prior user handle, including followers and following data, will be transferred to your new user handle.<p>@tvme
@T_V_
@TV123
@TommiV<p>Please reply to this email if you have any questions.<p>Best regards,
X<p>--8<--<p>Good thing they appreciate my loyalty! I appreciate how much of an asshole the person behind this decision is.
Upvote: | 57 |
Title: This is rather a mature project. It began out of curiosity: I wanted to test if Haskell FFI was powerful and expressive enough to interconnect C and Haskell code flawlessly. Particularly, if C strings generated inside Nginx can be shared within Haskell code and what must be done to respect their lifetimes etc.<p>Recently, I released version 3.2.0 with revamped README (with a lot of examples) and a new approach to building Haskell handlers using the modernized cabal v2-build for dependencies.
Upvote: | 98 |
Title: Hi!<p>It's a complete product with integrations to Auth0, OpenAI, Google Cloud and Stripe, which consists of Next.js Web App, Node.js + Express Web API and Python + FastAPI AI API<p>I've built this software, because I wanted to make money by selling tokens to enable users talking with the chatbot. But I think Google / Apple will include such AI-powered assistant in their products soon, so nobody will pay me for using it<p>So I open source the product today and share it as a GNU GPL-2 licensed software<p>I'm happy to assist in case if something is unclear or requires additional docs and answer any questions about Gdańsk AI :)<p>Thanks
Upvote: | 100 |
Title: I'm here to vent and give a heads-up to you all. It's about my nightmare with Discord as a game developer - the platform where I've been nurturing my game's community for the last 3 years.<p>Imagine building something from scratch, pouring your heart into it, watching it grow day by day. My server has over 4,000 die-hard players after 3 years, buzzing with ideas, feedback, and some crazy game theories. It was not just a server - it was my lifeline, my nerve center.<p>Now imagine waking up one morning to find my account disabled. Poof! Because someone out there decided to play a prank and report me for being "underage." I lodged an appeal ASAP, throwing in every proof of identity I had. You'd think that'd be enough, right? Well, guess again.<p>It's been over 3 days now, and Discord Support is as silent as a graveyard. Their policy says that my account will be dust after 14 days if things don't get sorted. Each tick of the clock is like a stab, knowing my community is sitting there in the dark. It's beyond frustrating, it's crushing. I'm unable to reach them because if I make an alt account to announce this, what's stopping some random strangers from making their own claim that they are the dev?<p>You might think, "Surely they'll respond before the deadline." And I wish I could believe that. But I've heard enough horror stories about Discord support responding when it's too late, or never responding at all, and the account just disappears into oblivion. That fear is real, folks.<p>So, here's my two cents. Seriously think over where you're setting up your community. Consider if you have any real control over the platform. This helplessness I'm feeling now, the feeling of seeing something you've worked on for years slipping away... It sucks, big time.<p>My advice? Don't put all your eggs in one basket, especially if that basket is Discord. I'm learning the hard way and trust me, it's a lesson you don't want to experience.<p>Any advice or similar horror stories are more than welcome. But more than anything, I hope my painful story can save someone else from this mess. Lastly, I'm not giving away my dev/game name because I heard that its possible for anyone to reach out to discord and claim my server once my disabled account gets deleted after 14 days.<p>Note: If there's a kind soul from Discord seeing this, I would really appreciate your help. My ticket # is 38363509.
Upvote: | 80 |
Title: It is fascinating that neural networks have such a run at the moment. I wonder if this will continue "forever". Or if we will see a different paradigm eclipse them in the future.<p>Is anybody still doing research in the area of genetic programming?<p>The genetic programming books of John R. Koza were the first I ever read about machine learning. It felt like magic at that time.<p>I have the feeling that the approach to generate programs for the CPU via evolution still has a lot to offer if it was explored further.<p>If there is research going on out there, I would love to follow it.
Upvote: | 51 |
Title: Name Checker is a tool for when you want to create a new project and don’t want to check all the places (npm, apt, GitHub) for a name collision, you can just use this tool instead.<p>I started working on this because I noticed on HN folks sometimes create a new company and SAAS and they didn’t notice that the name was already taken.<p>Please let me know if y'all have any questions or comments!
Upvote: | 62 |
Title: I’m seeing some products that transcribe songs, but I want to tell an AI to write the exercises I describe for practice purposes. For example: “Write the chromatic 7th intervals from C to B in eighth notes”
Upvote: | 44 |
Title: hey HN. this is a Language Server[0] designed specifically for Postgres. A language server adds features to IDEs (VSCode, NeoVim, etc) - features like auto-complete, go-to-definition, or documentation on hover, etc.<p>there have been previous attempts at adding Postgres support to code editors. usually these attempts implement a generic SQL parser and then offer various "flavours" of SQL.<p>This attempt is different because it uses the actual Postgres parser to do the heavy-lifting. This is done via libg_query, an excellent C library for accessing the PostgreSQL parser outside of the server. We feel this is a better approach because it gives developers 100% confidence in the parser, and it allows us to keep up with the rapid development of Postgres.<p>this is still in early development, and mostly useful for testers/collaborators. the majority of work is still ahead, but we've verified that the approach works. we're making it public now so that we can develop it in the open with input from the community.<p>a lot of the credit belongs to pganalyze[1] for their work on libpg_query, and to psteinroe (<a href="https://github.com/psteinroe">https://github.com/psteinroe</a>) who the creator and maintainer.<p>[0] LSP: <a href="https://microsoft.github.io/language-server-protocol/" rel="nofollow noreferrer">https://microsoft.github.io/language-server-protocol/</a><p>[1] pganalyze: <a href="https://pganalyze.com/" rel="nofollow noreferrer">https://pganalyze.com/</a>
Upvote: | 963 |
Title: On Friday there was a thread about "who's behind Archive.today?" [0], and it linked to a post by the developer saying [1]:<p><i>Archive.today is doomed to eventually die.</i><p>It's estimated the quantity of data under management is 1 petabyte. Given they've been going for around 8-10 years, we can extrapolate this is only 100-150 terabytes per year.<p>While not completely trivial, this is only 5 or 6 hard drives at today's maximum capacity of 28TB.<p>What if the developer published the data to a torrent on a monthly or other periodic basis? While not completely legal, it would enable their legacy to live on indefinitely and ensure future access to a lot of information.<p>[0] <i>Archive.today: on the trail of mysterious guerrilla archivists of the Internet</i> - https://news.ycombinator.com/item?id=37009598 - Aug 2023 (58 comments)<p>[1] <i>"archive.today is doomed to die"</i> https://blog.archive.today/post/657822115776610304/not-respecting-peoples-privacy-copyright-laws
Upvote: | 54 |
Title: I'd like to write a software stack from scratch, starting from low level assembly to explore alternative and novel avenues for computing.<p>I'm looking for some hardware that comes with as many modern features such as multiple cores, USB, PCI, etc., and with datasheets and open specs for all components of the board, that are reasonably possible for a single person to grok.<p>That cuts out all modern x86 and ARM systems (you need a lifetime just to program a single GPU driver), RISC-V boards still depend on binary blobs and underdocumented auxiliary chips.<p>Without going as far down to a Z80 microcomputer, what is the most powerful and programmable computer I can plug a keyboard and monitor to, and start hacking?
Upvote: | 44 |
Title: <p><pre><code> '' Recent line
`` Recent line and column
</code></pre>
(It's two single-quotes, not a double quote.)
Upvote: | 50 |
Title: I keep seeing critiques against working on tasks that are meant to maximize user engagement. A recent comment that I saw on the subject went in the lines of: <these are not the big/important problems of our lives>. Therefore my question comes because I cannot seem to see by myself what the real problems are.<p>A short argumentation would be valuable too, just so it's understandable where you come from.<p>Thanks & Happy Monday!
Upvote: | 122 |
Title: Could be a painting, or a boat. I'm interested in what HN'ers are building that isn't software-related.
Upvote: | 377 |
Title: Hi HN, I’m Hamza from GitStart (<a href="https://gitstart.com/">https://gitstart.com/</a>). We’re building a service for engineering teams to assign tickets and get back PRs. Here’s a video of how it works: <a href="https://www.loom.com/share/61d46a5b07e04b45a51af2a5008ee173" rel="nofollow noreferrer">https://www.loom.com/share/61d46a5b07e04b45a51af2a5008ee173</a><p>We recruit devs who want to grow from remote countries, to work full-time on these tickets and review each other’s PRs. For tech teams it’s pay as you go, and they only pay if they merge the PRs (we pay our devs a base regardless).
GitStart is very personal to me because I struggled myself to start my career as a junior dev. Being born in Pakistan made it super hard to find my first remote job. Things changed with Google Summer of Code, which was the first time I got paid to contribute to a large production codebase.<p>When I became a staff engineer, I started mentoring junior devs on their solo projects. I then realized what they really needed was production code experience, so I started sending them tickets from my backlog to grow them through code reviews. I also paid them for it. They learned a lot and I got a lot done, which started GitStart.<p>That’s why, in the description above, we say “devs who want to grow”. This is key to what we do—we are not just a body shop or middleman, and we don’t charge by the hour. We provide useful mentorship to junior devs. We want to be a meaningful part of their career path.<p>Our thesis (to use that slightly pretentious word) is that the economics work better this way as well. There is a lot of junior dev talent in these countries whose potential will bever be realized by hourly piecework. What’s needed are longer term relationships and work engagements, over the course of which a junior dev can learn and grow on a large production codebase. Much of this has to do with learning the culture of effective tech teams.<p>On the other side, effective tech teams don’t want hourly piecework either. Clients tell us they love having developers dedicated to their project, becoming more familiar with the code over time. This is the way for value to be maximized on both sides.<p>A number of commercial open source repos already use us, so you can check out some of the PRs created by GitStart here:<p>Cal.com: <a href="https://github.com/calcom/cal.com/pulls?q=is%3Apr+author%3Agitstart-calcom+is%3Aclosed">https://github.com/calcom/cal.com/pulls?q=is%3Apr+author%3Ag...</a><p>SourceGraph: <a href="https://github.com/sourcegraph/sourcegraph/pulls?q=is%3Apr+author%3Agitstart-sourcegraph+is%3Aclosed">https://github.com/sourcegraph/sourcegraph/pulls?q=is%3Apr+a...</a><p>StoryBooks: <a href="https://github.com/storybookjs/storybook/pulls?q=is%3Apr+author%3Agitstart+is%3Aclosed">https://github.com/storybookjs/storybook/pulls?q=is%3Apr+aut...</a><p>Strapi: <a href="https://github.com/strapi/strapi/pulls?q=is%3Aclosed+is%3Apr+author%3Agitstart">https://github.com/strapi/strapi/pulls?q=is%3Aclosed+is%3Apr...</a><p>Supabase: <a href="https://github.com/supabase/supabase/pulls?q=is%3Apr+author%3Agitstart+is%3Aclosed">https://github.com/supabase/supabase/pulls?q=is%3Apr+author%...</a><p>Twenty: <a href="https://github.com/twentyhq/twenty/pulls?q=is%3Apr+author%3Agitstart-twenty+is%3Aclosed">https://github.com/twentyhq/twenty/pulls?q=is%3Apr+author%3A...</a><p>Our main technical challenge has been securing code sharing. One solution was building GitSlice, which enables creating sub-repos that sync with the upstream repo. When GitStart devs create PRs on the platform, GitSlice syncs them upstream while pulling back CI/CD checks and review comments. This enables our devs to contribute with limited codebase access.<p>To prevent slices from breaking, we verify they run within a docker container, which also enables us to build review environments. Fun fact: we managed to support native iOS and Android codebases by building and running them on appetize.io instead of docker.<p>There have been countless attempts at this space so we would love to hear your feedback on how we approached this problem or your past experiences working with junior devs in this way. We look forward to a good conversation!
Upvote: | 235 |
Title: Hi HN, we are Nick and Ben, creators of Axilla - an open source TypeScript framework to develop LLM applications. It’s in the early stages but you can use it today: we’ve already published 2 modules and have more coming soon.<p>Ben and I met while working at Cruise on the ML platform for self-driving cars. We spent many years there and learned the hard way that shipping AI is not quite the same as shipping regular code. There are many parts of the ML lifecycle, e.g., mining, processing, and labeling data and training, evaluating, and deploying models. Although none of them are rocket science, most of the inefficiencies tend to come from integrating them together. At Cruise, we built an integrated framework that accelerated the speed of shipping models to the car by 80%.<p>With the explosion of generative AI, we are seeing software teams building applications and features with the same inefficiencies we experienced at Cruise.<p>This got us excited about building an opinionated, end-to-end platform. We started building in Python but quickly noticed that most of the teams we talked to weren’t using Python, but instead building in TypeScript. This is because most teams are not training their own models, but rather using foundational ones served by third parties over HTTP, like openAI, anthropic or even OSS ones from hugging face.<p>Because of this, we’ve decided to build Axilla as a TypeScript first library.<p>Our goal is to build a modular framework that can be adopted incrementally yet benefits from full integration. For example, the production responses coming from the LLM should be able to be sent — with all necessary metadata — to the eval module or the labeling tooling.<p>So far, we’ve shipped 2 modules, that are available to use today on npm:<p>* *axgen*: focused on RAG type workflows. Useful if you want to ingest data, get the embeddings, store it in a vector store and then do similarity search retrieval. It’s how you give LLMs memory or more context about private data sources.<p>* *axeval*: a lightweight evaluation library, that feels like jest (so, like unit tests). In our experience, evaluation should be really easy to setup, to encourage continuous quality monitoring, and slowly build ground truth datasets of edge cases that can be used for regression testing, and fine-tuning.<p>We are working on a serving module and a data processing one next and would love to hear what functionality you need us to prioritize!<p>We built an open-source demo UI for you to discover the framework more: <a href="https://github.com/axilla-io/demo-ui">https://github.com/axilla-io/demo-ui</a><p>And here's a video of Nicholas walking through the UI that gives an idea of what axgen can do: <a href="https://www.loom.com/share/458f9b6679b740f0a5c78a33fffee3dc" rel="nofollow noreferrer">https://www.loom.com/share/458f9b6679b740f0a5c78a33fffee3dc</a><p>We’d love to hear your feedback on the framework, you can let us know here, create an issue on the GitHub repo or send me an email at [email protected]<p>And of course, contributions welcome!
Upvote: | 161 |
Title: Hi HN - Today we’re launching V2 of Phind.com, an assistant for programmers that combines GPT-4 with the ability to search the web and your codebase to solve nearly any technical problem – no matter how difficult.<p>We’re incredibly grateful for the feedback we received when we first launched GPT-4 answers back in April (<a href="https://news.ycombinator.com/item?id=35543668">https://news.ycombinator.com/item?id=35543668</a>). As Phind has gotten better at complex programming tasks, the questions it gets asked have gotten more complex as well. In the past, we would always perform a web search for every input. This limitation constrained Phind’s answers to what was present in the search results, preventing us from making Phind a more powerful debugger and making it challenging to integrate Phind with your codebase.<p>We’ve addressed all these shortcomings in Phind V2. This release has three major updates: (1) Phind is now a pair programming agent that knows when to browse the web, ask clarifying questions, and call itself recursively; (2) the answering engine defaults to GPT-4, and you can use it without a login; (3) we integrate with your codebase via our new VS Code extension.<p>We realized that search is only one of many tools that Phind should be able to use. As such, Phind has been re-engineered to be an agent that can dynamically choose whatever tool best helps the user – it’s now smart enough to decide when to search and when to enter a specialized debug mode. Instead of making assumptions about your code and proceeding blindly, Phind can ask you questions and clarify its assumptions. When a problem requires multiple searches or logical steps to solve, Phind can call itself recursively and perform multi-step reasoning without user input.<p>We’ve heard from you that switching between your IDE and Phind in the browser has been a major pain point. No longer – we’re launching a VS Code extension that brings Phind into the IDE and finally connects Phind with the context of your codebase. Phind in VS Code automatically determines which parts of your code are relevant to your search and can help you squash bugs in a single click.<p>To maximize Phind’s alignment with your preferred answer style, we’ve also added a feature called Answer Profile where you can tell the AI about yourself. Phind will apply this answering style across the board, automatically.<p>Here are some examples of the new Phind answering questions it could not before:<p>Clarifying assumptions to help a user with debugging: <a href="https://www.phind.com/agent?cache=cljmjzjgn0000jo085otq111f">https://www.phind.com/agent?cache=cljmjzjgn0000jo085otq111f</a><p>Designing a highly specific and custom database schema: <a href="https://www.phind.com/agent?cache=clkwpprz600g4jt08dl21e7r6">https://www.phind.com/agent?cache=clkwpprz600g4jt08dl21e7r6</a><p>Splitting a Wordpress theme across multiple files: <a href="https://www.phind.com/agent?cache=clknqywuq001pji083sdacf9p">https://www.phind.com/agent?cache=clknqywuq001pji083sdacf9p</a><p>Phind’s asking clarification questions in debug mode: <a href="https://www.phind.com/agent?cache=cljmjzjgn0000jo085otq111f">https://www.phind.com/agent?cache=cljmjzjgn0000jo085otq111f</a><p>Phind extension answering questions about a local codebase: <a href="https://www.phind.com/agent?cache=ra4kh2v3epgv5iw7z6dlzuo4">https://www.phind.com/agent?cache=ra4kh2v3epgv5iw7z6dlzuo4</a><p>Answering questions about a local codebase using the web: <a href="https://www.phind.com/agent?cache=ztiaju6xwtpi39l2kjdnwh20">https://www.phind.com/agent?cache=ztiaju6xwtpi39l2kjdnwh20</a><p>We are incredibly grateful for the feedback the HN community has given us and are excited to hear your thoughts about this release!<p>Cheers,
Michael and Justin
Upvote: | 221 |
Title: unfortunately, all online meetings in our institution are over Zoom. I never trusted the company for obvious reasons. but given their new ToS, I'm looking for alternatives that respect privacy.<p>Any alts must enable desktop sharing and must be easy to use by old people.
Upvote: | 47 |
Title: Hi HN! We’re been working hard on this low-code tool for rapid prompt discovery, robustness testing and LLM evaluation. We’ve just released documentation to help new users learn how to use it and what it can already do. Let us know what you think! :)
Upvote: | 177 |
Title: Hi!<p>As I was working on a side project, I noticed I wanted to use SQLite like a Document Database on the server.
So I built Doculite. DocuLite lets you use SQLite like Firebase Firestore. It's written in Typescript and an adapter on top of sqlite3 and sqlite.<p>Reasons:<p>1) Using an SQL Database meant having less flexibility and iterating slower.<p>2) Alternative, proven Document Databases only offered client/server support.<p>3) No network. Having SQLite server-side next to the application is extremely fast.<p>4) Replicating Firestore's API makes it easy to use.<p>5) Listeners and real-time updates enhance UX greatly.<p>6) SQLite is a proven, stable, and well-liked standard. And, apparently one of the most deployed software modules right now. (src: <a href="https://www.sqlite.org/mostdeployed.html" rel="nofollow noreferrer">https://www.sqlite.org/mostdeployed.html</a>)<p>What do you think? Feel free to comment with questions, remarks, and thoughts.<p>Happy to hear them.<p>Thanks
Upvote: | 160 |
Title: My dad died a few years back due to unexpected circumstances so wasn't able to talk to him beforehand which made digging through his digital stuff difficult until I stumbled upon a text file on his PC with all his U/Ns and PWs. Of course thats not the best way to do it but it sure was super helpful. Bunch of stuff I didn't even think about (airline miles, Dropbox's with pictures, interesting forum content, memberships, etc. 50+ accounts total).<p>What's the better more secure way to store this info along with any wishes regarding your digital (after)life if I want to make it easier for my kids or whomever digs through it later? Any additional info thats helpful to include? Who to entrust it with?
Upvote: | 42 |
Title: Mine is a jife, a knife specifically designed for peanut butter. My wife was bemused when I purchased it, but she uses it every day now.
Upvote: | 55 |
Title: Hi HN - We’re Ed and George, the team behind Stellar Sleep (<a href="https://stellarsleep.com/">https://stellarsleep.com/</a>). We spent the past 1.5 years developing an app that helps people break the insomnia cycle using psychology.<p>Insomnia has been a personal struggle for both of us. George’s insomnia led to him quantifying his sleep for the last 7 years and making questionable life choices like wearing orange (blue-light blocking) goggles to evening social events.<p>I (Ed) suffered from chronic insomnia for over a year and tried everything from cutting coffee, blocking blue light, to taking melatonin and antihistamine, but couldn’t find anything that worked. It took an immense amount of energy for me to perform even the most mundane of tasks, such as doing my laundry or ordering groceries. At night, I felt overwhelming loneliness and resentment as I lay in bed wide awake, mindlessly scrolling through reddit and browsing HN in the middle of the night. I even bought a $500 research-grade EEG device to track my sleep, which was honestly kind of depressing because it showed that I was sleeping less than 4 hours per night for weeks straight.<p>When I finally decided to see a sleep specialist, I was put on a 3-month long waiting list. Eventually, I was able to get my insomnia treated. That finally helped. But it also made me realize that the same therapy that I received could be delivered online—there is no reason why anyone should have to wait 3 months to get treatment. George and I both have experience in digital health, so we decided to partner with sleep experts to create Stellar Sleep.<p>People with chronic insomnia often receive inadequate care advice. They’re told to try cutting back on coffee, meditate more, or make their bedroom darker, when chronic insomnia is often rooted in deeper, psychological issues (e.g., financial anxiety, workplace stress…etc.). In fact, “<i>present research [now] supports the view that sleep is causally related to mental health difficulties</i>” (Sleep Medicine Reviews, Dec 2021, doi: 10.1016/j.smrv.2021.101556). This matches up with what we observe—over 50% of our users reported that they have been diagnosed with depression, anxiety or other mental health conditions. Stellar Sleep uses psychology to help chronic insomniacs get their lives back.<p>We start by understanding a person’s immediate sleep-related challenges and providing them with practical strategies for dealing with them. Then, we drill down on the underlying psychological drivers of a person’s chronic insomnia and create a program to resolve those drivers one by one. Participants spend 5-10 minutes per day completing modules in the Stellar Sleep app.<p>We launched last August, and have already helped over 10,000 patients improve their sleep permanently. Here’s how one of our customers, Victoria, describes her experience:<p>“<i>I’ve had insomnia for about 5 years. I was at the point of desperation, willing to try anything. Insomnia feels very lonely. It’s 2am and your spouse is asleep. You’re very alone with your anxious thoughts. About the mortgage rates changing. About all the things you have to do tomorrow without any sleep. It feels like you’re trapped and alone.<p>After starting Stellar Sleep I’ve now started sleeping better for the first time in 5 years. I’ve gotten hope back that I can be normal. Most people who sleep well don’t worry about sleep. A good night’s sleep should just be normal.</i>”<p>We have strong clinical data showing that we're just as effective as in-person treatment and have a year-long clinical study running with Harvard Medical School / Brigham & Women’s Hospital paid for by the State of Massachusetts and the US Department of Commerce. On average, our users sleep 74 minutes longer than before and spend 52% less time awake in the middle of the night.<p>If you have trouble with sleep, try our app and let us know what you think! If you have any questions, please feel free to ask and we will be happy to answer them. We look forward to hearing your comments and experiences!<p>P.S. Here's our FAQ page (<a href="https://stellarsleep.com/faqs">https://stellarsleep.com/faqs</a>) since so many of you asked about pricing
Upvote: | 127 |
Title: Hi HN,<p>A few of our team members at Airbyte (and Joe, who killed it!) recently played with building our own internal support chat bot, using Airbyte, Langchain, Pinecone and OpenAI, that would answer any questions we ask when developing a new connector on Airbyte.<p>As we prototyped it, we realized that it could be applied for many other use cases and sources of data, so... we created a tutorial that other community members can leverage [<a href="http://airbyte.com/tutorials/chat-with-your-data-using-openai-pinecone-airbyte-and-langchain">http://airbyte.com/tutorials/chat-with-your-data-using-opena...</a>] and the Github repo to run it [<a href="https://github.com/airbytehq/tutorial-connector-dev-bot">https://github.com/airbytehq/tutorial-connector-dev-bot</a>]<p>The tutorial shows:
- How to extract unstructured data from a variety of sources using Airbyte Open Source
- How to load data into a vector database (here Pinecone), preparing the data for LLM usage along the way
- How to integrate a vector database into ChatGPT to ask questions about your proprietary data<p>I hope some of it is useful, and would love your feedback!
Upvote: | 220 |
Title: Hi all, new to posting but not to the forum. But, I’m struggling with my career, where to do next, and I’m throwing a line out for anyone that may have their own lessons learned to share.<p>For context, I’m a project manager in the EV charging infrastructure space. Nothing to do with software, as I’m primarily involved with the physical installation. I’ve been a PM for the better part of a decade in different industries and find myself a bit stuck in the ‘crisis of meaning’ stage.<p>I’m good at my job, but it’s not fulfilling and although I’m senior level I still haven’t broken through the $100k ceiling. Though, I do live in a medium cost of living area.<p>So, what am I asking? Well, I want to shift over to a position that will allow higher earnings and decent personal growth. Maybe it’s a unicorn job to get money and fulfillment, but the search continues. Given the housing crisis and crazy prices, I just feel I need to do what I can to increase earnings. Like most people, I just want to comfortably afford a modest house.<p>So, to all the high earners out there, do have any general advice? Any top of mind jobs I could chase, given my PM background?<p>I realize how broad this post is, but I’m desperate for a thread to follow and an outside perspective. Thanks
Upvote: | 56 |
Title: So, it feels like this should exist. But I couldn't find it. So I tried to build it.<p>Agentflow lets you run complex LLM workflows from a simple JSON file. This can be as little as a list of tasks. Tasks can include variables, so you can reuse workflows for different outputs by providing different variable values. They can also include custom functions, so you can go beyond text generation to do anything you want to write a function for.<p>Someone might say: "Why not just use ChatGPT?" Among other reasons, I'd say that you can't template a workflow with ChatGPT, trigger it with different variable values, easily add in custom functions, or force the use of custom functions for steps in the workflow.<p>Someone might also say: "Then why not use Auto-GPT or BabyAGI?" Among other reasons, I'd say you can't if you want consistency because these tools operate autonomously, creating and executing their own tasks. Agentflow, on the other and, lets you define a step-by-step workflow to give you more control.<p>I'd like to do more with this, including adding more custom functions, and more examples, and more ways to trigger workflows (such as in response to events). But first, I want to make sure I'm not wasting my time! For starters, if something like this already exists, please tell me.
Upvote: | 51 |
Title: Hi HN,<p>If you’re selling templates or digital assets online, platforms like Gumroad have a ton of amazing features . . . but they’re also expensive. It’s not uncommon to be paying 10%, 20% or even 30% of your revenue just to host and deliver some digital content to customers.<p>Instead, we think most creators should own their own Stripe account and use a lightweight fulfillment layer to send customers their orders.<p>So we built Easyful, a platform built on Stripe to email your content to customers when they buy it. And it’s free!<p>We’ve been using Easyful ourselves for a few months now. Try it out and let us know what you think!
Upvote: | 159 |
Title: Hi, we are Ali, Hassan, and Alistair, co-founders of Infracost (<a href="https://www.infracost.io/">https://www.infracost.io/</a>). Infracost helps engineers see the cost of each Terraform change before launching resources. When changes are made, it posts a comment with the cloud cost impact. For example, “you’ve added 2 instances and volumes, and change an instance type from medium to large, your bill will increase by 25% next month, from $1000 to $1250 per month”.<p>We launched in February 2021 (<a href="https://news.ycombinator.com/item?id=26064588">https://news.ycombinator.com/item?id=26064588</a>), and Infracost is now being actively used by over 3,000 companies. However, there is a shift happening in the cloud cost management space. New teams, called FinOps teams (a combination of "Finance" and "DevOps"), are being formed within companies to manage cloud costs.<p>One of the first tasks assigned to these teams is to determine "who is using what" - that is, which teams, business units, products, etc. are spending the most on cloud. To accomplish this, they use tags. Tags are labels that all cloud resources should have and are key-value pairs. For example, a server could be tagged with: product=HackerNews; environment=production; team=blueTeam. So if resources are not tagged properly, then you can’t tell who is using what.<p>However, FinOps teams face challenges because their tools are reactive. These tools begin by analyzing cloud bills and providing visibility of tags from there. This means that they are looking at resources that are already running in production and costing money. A customer recently shared, “I want all resources to be properly tagged. But if they are not, I would rather a resource not be tagged at all than be tagged incorrectly.”<p>My "aha" moment! FinOps teams can define a tagging policy that can be validated in CI/CD before resources are launched. This is important because if code is shipped with the wrong tags, FinOps teams will have to fight for sprint time to fix them. Even if you shut down an untagged resource directly in the cloud, the next time Terraform runs, the resource will launch again with no tag. You need to fix the issue at its root.<p>I’d love your feedback on our solution to the tagging problem. You define your tag key-value policy in our SaaS product, and Infracost checks all Terraform resources per change. If anything fails the policy, it posts a comment with the details of which resources need tags, and what the allowed values are. Once fixed, it will let the code be shipped to production.<p>Try it out by going to <a href="https://dashboard.infracost.io/">https://dashboard.infracost.io/</a>, setting up with the GitHub app or GitLab app, and defining your tagging policy. It will then scan your repository and inform you of any missing tags and their file and line number. You can use the free trial, but if you need more time, please message me and I’ll extend it for you.<p>I would also love to hear how others ensure that the correct tag keys and values are applied to all resources, and whether this is done proactively or reactively. Additionally, I would be interested in hearing about any lessons learned in the process.<p>Cheers
Upvote: | 86 |
Title: Hey HN!<p>You know when you wanna build a new PC, and need to watch a bunch of videos to find the best parts, and then tweak your shopping cart to fit your budget? Well, this new app can help you.<p>It recommends the best hardware for your usage (gaming or work), and considers your budget. It's really helpful, even when you got expertise on hardware. Check it out, and feel free to suggest stuff! Hope you like it!<p><a href="https://www.pcbuilderai.com/" rel="nofollow noreferrer">https://www.pcbuilderai.com/</a>
Upvote: | 47 |
Title: After reading about what happened with NightOwl yesterday [0], I thought about what it would take to be aware of things like that in the future. I created ToSNotify to automatically notify you when a website's terms change.<p>A harder problem I've been thinking through is how to know which terms to track, since it'd be a pain to add every site I have an account with. One idea I had is to automatically get terms for apps you have installed from the app store. Any other ideas/feedback are appreciated!<p>[0]: <a href="https://news.ycombinator.com/item?id=37052508">https://news.ycombinator.com/item?id=37052508</a>
Upvote: | 190 |
Title: Hello HN community,<p>I wanted to create a space for anyone interested in making new friends. Whether you're open to online or in-person hangouts, feel free to drop a comment below. It's a chance to expand our networks and connect with like-minded folks.<p>To start off, here are a few of my interests: computational biology, game design, nature photography. Open to online friends.
You can reach me on Discord where my handle is hyperific.<p>Looking forward to reading your intros and interests.
Upvote: | 50 |
Title: Hi HN! We’re the co-founders of Refine (<a href="https://github.com/refinedev/refine">https://github.com/refinedev/refine</a>), an open-source platform for developing enterprise web applications rapidly. In one year since launch we have 15K active developers each month, and over 5K projects in production. Among the 5,000+ projects deployed to production, we see a lot of admin panels, dashboards, B2B portals and SAAS interfaces.<p>Check out our online generator at (<a href="https://refine.dev/#playground">https://refine.dev/#playground</a>) to create a custom Refine application and download it.<p>Before starting refine, we spent five years as a consultancy company building internal-facing applications for enterprise clients. We have seen many complex use cases where the demanded flexibility was much higher than what existing low-code/no-code solutions provide. Moving away from rigid architectures and pre-made components that are difficult to customize, developers tend to start from the scratch. This results in a significant waste of time and resources.<p>In order to find a sweet spot between “starting from scratch” and higher level solutions, we started working on refine. After a couple of iterations, we came up with a “headless” solution, separating the UI-layer completely from the rest of the frontend logic. This allowed us to support multiple UI frameworks and custom designs out of the box.<p>We understand the importance of frontend developers working with the stack and tools they love and are familiar with. So, we have built our project creation wizard that allows developers to mix and match their technology stack when creating a project.<p>After 1.5 years of development with the great support of the open-source community, Refine has become a mature framework that allows developers to rapidly build enterprise applications and have 100% control over their projects.<p>While remaining unobtrusive, Refine still saves a lot of development resources by eliminating repetitive tasks such as CRUD operations, state management, routing, authentication, access control, and i18n. It’s also backend-agnostic and works with any APIs or services.<p>The best way to start Refine development is visiting our extensive documentation at (<a href="https://refine.dev/docs/">https://refine.dev/docs/</a>). Here, you can also find tutorials and real-life examples to use as a starting point for your use case.<p>We had one successful Show HN earlier this year (<a href="https://news.ycombinator.com/item?id=34515128">https://news.ycombinator.com/item?id=34515128</a>), and have since had several major releases and added many more features.<p>Recently, we've released our enterprise edition featuring deployments, built-in security, backend integrations and auto-generated UI's. By expanding Refine's capabilities beyond the frontend, the enterprise edition offers a complete solution for the internal tooling requirements of larger organizations. To get more information, please see our pricing page: <a href="https://refine.dev/pricing/">https://refine.dev/pricing/</a>.<p>If you have any questions, ideas or suggestions please share them with us. The team will be here all day to answer. We look forward to all of your comments!
Upvote: | 259 |
Title: Was wondering if I was to buy cheapest hardware (eg PC) to run for personal use at reasonable speed llama 2 70b what would that hardware be? Any experience or recommendations?
Upvote: | 70 |
Title: i'd forgotten i'd written this a year and a half ago, and when a friend passed me the link to it just now, it seemed hilarious to me. probably some other people will enjoy it too
Upvote: | 47 |
Title: We have gotten over 1000 voice messages left by the users of our platform.<p>We take privacy seriously, so all data are anonymized and are not sold to anyone.<p>So far, we had a user who said that 'had it not been for Bubblic, I might not be here today'. This gives us so much drive to carry on with our project!<p>We'd appreciate any feedback you have :)
Upvote: | 68 |
Title: Maybe it is a life hack, a new topic, a new recipe, etc. Whatever YOU think is exciting to you.
Upvote: | 47 |
Title: Hey HN! We’re Jeremy, Abhi and Darren from Rubbrband (<a href="https://www.rubbrband.com/">https://www.rubbrband.com/</a>). Rubbrband is software that evaluates the quality of images generated by Stable Diffusion (or Midjourney, DALL-e, etc…).<p>We actually stumbled into this idea while working on solving a different problem. While helping a few companies use Stable Diffusion in production, we found that these companies needed to use QA analysts to manually evaluate their images to make sure they were high quality once in prod. This process often took days just to go through thousands of images, so companies that have a text-to-image product with PMF couldn’t guarantee that their images were of high quality.<p>From our initial set of customers, we found that evaluating the quality of their outputs at scale was an even larger problem than just using the model itself. So we pivoted our company into working on solving this problem using our skills from CV research. All three of us did computer vision research at UC Berkeley, and Abhi worked with John Canny, who invented many CV techniques like the Canny Edge Detector.<p>We built a product that automates this process using computer vision. We’ve trained in-house several computer vision models that grade images based on different criteria.<p>These include: - Detecting human deformities, such as a person with 7 fingers on a hand (image generation models generate deformed hands over 80% of the time when generating a photo of a person! This includes the state of the art: Midjourney, SDXL, Runway, etc…); - A score that rates how well the image aligns with the prompt(we do this using our own finetuned Visual Question Answer model); - A composition score (how well composed the image is according to photography “rules”)<p>Using Rubbrband is pretty simple. You can send an image to us to process via our API or our web app, and you’ll get scores for each of those criteria back on your dashboard in less than 10 seconds. Here is a quick Loom demo: <a href="https://www.loom.com/share/961830347b3643dcbb92dfe80f8ca1f0?sid=faf1d9fb-5023-4dc8-96fc-5b3806ef3c6b" rel="nofollow noreferrer">https://www.loom.com/share/961830347b3643dcbb92dfe80f8ca1f0?...</a>.<p>You can filter your images based on certain criteria from the dashboard. For instance, if you want to see all images with deformed eyes, you can click the “deformed eyes” filter at the top of the screen to see all of those images.<p>We store images generated by your image generation model. We’re like a logging tool for images, with evaluations on top. We currently charge $0.01 per image, with your first 1000 images free.<p>We’re super excited to launch this to Hacker News. We’d love to hear what you think. Thanks in advance!
Upvote: | 104 |
Title: Hey HN! We're Phil and Ming, co-founders of Retake (<a href="https://github.com/getretake/retake">https://github.com/getretake/retake</a>). Retake is an open source tool that adds keyword and semantic (i.e hybrid) search to databases. We’ve started by extending the capabilities of Postgres with an SDK for lightning-fast queries.<p>We built Retake to fix two issues: keeping vectors in sync with Postgres in real time is difficult, and most vector databases aren’t built for hybrid search.<p>A quick refresher: “keyword search” refers to a technique where results are scored based on the appearance of exact words or terms. “Semantic search” uses vector embeddings to understand the meaning behind those words. Hybrid search combines these two approaches to enhance the precision and relevance of results.<p>To implement semantic or hybrid search today, most organizations run batch jobs that update their search engine or vector database using ETL tools or custom data pipelines. We’ve seen from firsthand experience how time-consuming and costly this can be, as moving vectors often requires re-embedding the entire data source.<p>We’ve also seen how many vector databases lack crucial features of “traditional” search: keyword-based (BM25) search, faceting/aggregations, highlighting, efficient filtering, etc.<p>Here’s how Retake works - our core is built on top of OpenSearch, which acts as a search engine and vector database. We leverage logical-replication-based Change Data Capture (CDC) to stay in sync with Postgres, so documents and vectors are updated incrementally and in real time. Finally, Python and Typescript SDKs make it easy to integrate Retake into your application. There’s no need to manage separate vector databases and search engines, upload and embed documents, or run expensive reindexing jobs. All you need to think about is writing search queries.<p>The easiest way to get started with Retake is by running our Docker Compose stack:<p><pre><code> git clone https://github.com/getretake/retake.git
cd retake/docker && docker compose up
</code></pre>
Retake is Apache licensed and our repo is here: <a href="https://github.com/getretake/retake">https://github.com/getretake/retake</a>. For next steps, see our quick start guide: <a href="https://docs.getretake.com/quickstart">https://docs.getretake.com/quickstart</a><p>We’d love your feedback on our solution to hybrid search. Our focus right now is on nailing the basics, but we’d also love to hear what you think we should focus on next.
Upvote: | 88 |
Title: OFRAK Tetris is a project I started at work about two weeks ago. It's a web-based game that works on desktop and mobile. I made it for my company to bring to events like DEF CON, and to promote our binary analysis and patching framework called OFRAK.<p>In the game, 32-bit, little-endian ARM assembly instructions fall, and you can modify the operands before executing them on a CPU emulator. There are two segments mapped – one for instructions, and one for data (though both have read, write, and execute permissions). Your score is a four byte signed integer stored at the virtual address pointed to by the R12 register, and the goal is to use the instructions that fall to make the score value in memory as high as possible. When it's game over, you can download your game as an ELF to relive the glory in GDB on your favorite ARM device.<p>The CPU emulator is a version of Unicorn (<a href="https://www.unicorn-engine.org/" rel="nofollow noreferrer">https://www.unicorn-engine.org/</a>) that has been cross-compiled to WebAssembly (<a href="https://alexaltea.github.io/unicorn.js/" rel="nofollow noreferrer">https://alexaltea.github.io/unicorn.js/</a>), so everything on the page runs in the browser without the need for any complicated infrastructure on the back end.<p>Since I've only been working on this for a short period of time leading up to its debut at DEF CON, there are still many more features I'd eventually like to implement. These include adding support for other ISAs besides ARM, adding an instruction reference manual, and lots of little cleanups, bug fixes, and adjustments.<p>My highest score is 509,644,979, but my average is about 131,378.<p>I look forward to feedback, bug reports, feature requests, and strategy discussions!
Upvote: | 472 |
Title: deno_python 0.3.1 adds support for importing python pip packages directly in JavaScript! Fun and useful, slightly cursed.
Upvote: | 192 |
Title: An ex coworker (we are/were SWEs at Google) quit his job at the beginning of the year and started a solo SaaS solving a small and interesting niche b2b problem with generative AI, and his revenues are already $500k YTD (mostly profits, his hosting and marketing expenses are minimal). I am close to him and he showed me his Stripe account. No cofounders, no investors. Just himself, by design.<p>What an amazing dream, not having to deal with coworkers/managers/meetings/scrum/stand ups/reporting/RTO/etc. He spends his days doing programming, some customer support, self marketing on niche forums and subreddits, and thinking about the next idea.<p>It is truly my ideal life. I have had these thoughts for years, and every time I suppress them thinking “the grass is always greener, you’re a highly paid FAANG engineer, shut up and milk it”, but they just keep coming up in my head and I can’t shake off the feeling that I’m wasting my time in corporate America. It’s not even about the money, I am (currently) making more than my friend, it’s about the freedom. It would still be worth it even if it resulted in a 10x paycut for me.<p>How do I get an idea to start? I do not have any niche skills, I’m your typical run-of-the-mill backend developer with no particular expertise. Lately I’ve been learning the theoretical aspect of ML, and find it fascinating.<p>A couple other notes to hopefully prevent some of the most common responses:<p>- I am not afraid of hard work: I have worked in 80h a week hedge funds, and in no way I think of a solo business as “working less”. I am ready for it to take a large portion of my time and mental energy.<p>- I am fine with taking risk: I have worked in several early stage startups as an early employee, taking on many hats compared to my SWE role. I’ve done tech marketing, customer support, sales engineering, etc.<p>- I am in a financially stable position: I have multiple decades of living expenses because I aggressively saved up and invested in liquid assets ($4.5M in Vanguard index funds), no kids to support and no debt
Upvote: | 43 |
Title: Hi HN! We’re Dimitris and Youssef, founders of Wondercraft (<a href="https://www.wondercraft.ai/">https://www.wondercraft.ai/</a>), a platform that leverages AI voices to make podcast creation simple. This video shows how it works: <a href="https://www.loom.com/share/fa8ac8eba8b9440dbe0321ccb8ba9426?sid=fef9c08a-1003-4377-9d79-8d8a7bf14eb0" rel="nofollow noreferrer">https://www.loom.com/share/fa8ac8eba8b9440dbe0321ccb8ba9426?...</a>.<p>“Hacker News Recap” (<a href="https://www.wondercraft.ai/podcasts/hacker-news-recap">https://www.wondercraft.ai/podcasts/hacker-news-recap</a>) a podcast produced using our platform, has been running for 4 months and currently gets close to 23k listens per month. We’ve made its analytics publicly available: <a href="https://op3.dev/show/f77aea62-97e5-5cce-92c6-9464e51c30c6" rel="nofollow noreferrer">https://op3.dev/show/f77aea62-97e5-5cce-92c6-9464e51c30c6</a>.<p>Having previously attempted to start a podcast, we were well aware of the difficulties. Figuring out what equipment and software you need to buy is a daunting start. Editing is a lengthy and tedious process, technical difficulties often occur during recording, and planning logistics around recording is a hassle. As a result, content release is infrequent, which leads to lackluster growth.<p>At the same time, podcast consumption is experiencing exponential growth. There are 500M podcast listeners around the world, double in size compared to 5 years ago. Apart from the growth in listeners, podcasts are the medium that is most likely to influence behavior, which is the reason why the number of businesses having podcasts has grown 5x over the past 5 years. Finally, the last piece that led to the creation of Wondercraft is that text-to-speech models saw a big improvement about 6 months ago, with ElevenLabs releasing models with an output that is almost indistinguishable to humans (see HN thread here: <a href="https://news.ycombinator.com/item?id=34361651">https://news.ycombinator.com/item?id=34361651</a>).<p>Wondercraft integrates realistic text-to-speech with an infrastructure that simplifies podcast creation. For example, you can integrate music, publish your podcast / create an RSS feed, generate a video for your episode, get assistance in the script generation, auto generate show notes and transcript and translate your podcast all together. All text based tasks (e.g. script assistance, show note generation, etc) are completed using a chain of custom prompts to LLM models. All text-to-speech is done through custom voices that are either synthetically generated or professionally cloned from Voice Actors, using the ElevenLabs platform. Tasks such as episode translation involve the use of both LLMs and ElevenLabs. Video generation runs using Remotion and the RSS feed is an XML creation and updating routine.<p>Since launching, we’ve had more than 13k users sign up to create their podcast. Use cases that we’re seeing include: businesses repurposing their blogs and generating video content for their socials; writers/bloggers/newsletters reaching audience through another medium; news outlets and publications adding a news rundown podcast in their lineup; businesses creating internal educational/cultural material; and podcast studios using Wondercraft to serve client needs faster.<p>Wondercraft is not a tool for fully AI generated content. Rather, we save people time by transferring content they’ve created (e.g. an article they’ve written) to another medium. This technology is best suited for news rundowns and narrational format podcasts (often used by businesses talking about a niche topic). And while interview and conversational formats will sound better person-to-person, the logistical and (often) sound quality issues remain, so we’re testing out an “Async Podcasts” feature, where an interviewee can respond to questions async in writing, share a photo and (optionally) a clip of their voice, and a podcast will be created out of it.<p>We’d love to hear any thoughts, comments or experiences you may have had in relation to leveraging text to speech for podcast creation. Thank you for taking the time to read!
Upvote: | 153 |
Title: Hi HN,<p>pykoi is an open-source python library for ML scientists. pykoi makes it easier to collect data for LLMs, to use that data for finetuning, and to compare models to each other (e.g. your model pre- and post- finetuning, or your model vs openai vs claude). The library comes from pain points we experienced in LLM development:<p>1. Collecting feedback data from users isn't as easy as it could be. (The current process usually involves sharing excel files of annotated responses back-and-forth, offering no insight into how users actually engage with your models).<p>2. RLHF remains complicated to carry out. By <i>complicated</i>, we mean requires a lot of steps, hundreds of configs, lengthy setups, etc.<p>3. Comparing models to each other <i>as they're used</i> (that is, independent from academic metrics) is full of friction. The current approach: spin up a model, ask questions, write them down. Repeat for other models then compare.<p>At a high-level, we think that the active learning process should be closed-loop: data collection, fine tuning, and inference all feed from the same system. This library is our first step in that direction.<p>The project is still very early but we hope that some if it is useful. Note, we're fully open-source, and actively adding features!<p>Website: <a href="https://www.cambioml.com/pykoi">https://www.cambioml.com/pykoi</a>
GitHub: <a href="https://github.com/CambioML/pykoi">https://github.com/CambioML/pykoi</a><p>We would love your feedback!
Upvote: | 119 |
Title: I have a 2 year old son and I wonder how I will manage once he goes to school. He will inevitably envy the cool kids already plugged at all times on cellphones. Maybe this will cause social isolation?<p>For now, I try to give him a mostly screen free childhood, with occasional "treats" of dinosaur documentaries on rainy days.<p>How do you manage?
Upvote: | 69 |
Title: Hello hacker news,<p>I’m the maintainer of liteLLM() - package to simplify input/output to OpenAI, Azure, Cohere, Anthropic, Hugging face API Endpoints: <a href="https://github.com/BerriAI/litellm/">https://github.com/BerriAI/litellm/</a><p>We’re open sourcing our implementation of liteLLM proxy: <a href="https://github.com/BerriAI/litellm/blob/main/cookbook/proxy-server/readme.md">https://github.com/BerriAI/litellm/blob/main/cookbook/proxy-...</a><p>TLDR: It has one API endpoint /chat/completions and standardizes input/output for 50+ LLM models + handles logging, error tracking, caching, streaming<p>What can liteLLM proxy do?
- It’s a central place to manage all LLM provider integrations<p>- Consistent Input/Output Format
- Call all models using the OpenAI format: completion(model, messages)
- Text responses will always be available at ['choices'][0]['message']['content']<p>- Error Handling Using Model Fallbacks (if GPT-4 fails, try llama2)<p>- Logging - Log Requests, Responses and Errors to Supabase, Posthog, Mixpanel, Sentry, Helicone<p>- Token Usage & Spend - Track Input + Completion tokens used + Spend/model<p>- Caching - Implementation of Semantic Caching<p>- Streaming & Async Support - Return generators to stream text responses<p>You can deploy liteLLM to your own infrastructure using Railway, GCP, AWS, Azure<p>Happy completion() !
Upvote: | 140 |
Title: Recently I decided to migrate one of my side projects from AWS to Render for convenience and to manage some of my high ALB costs on AWS.<p>This weekend I began the process to move some of my infra to Render, including a static website, Postgres database and Redis. I switched the static website over but had some issues restoring my database from SQL backups (Render's Postgres appears to timeout connections after a certain amount of time).<p>After spending a few hours attempting to import all my data, I was logged out from dashboard and all my services were abruptly taken offline with no notice.<p>When trying to login again, it just throws me back to the login screen and sleuthing the network requests I can see I am now getting "Unauthorized" responses.<p>My production website was abruptly taken offline with the message "This service has been suspended by its owner." (I didn't suspend it)<p>For clarity, I was on a paid plan, had a credit card added, and all of these services were paid. I have received no communication, now I have to migrate back to AWS.<p>I can only assume they saw I was using the database heavily (normal during a restore) and decided I was abusing something? I absolutely did not violate any of their terms nor was I doing anything shady. It's just a static website and a Node service. Nothing crazy.<p>I just wanted to provide a warning to others because this seems egregious.
Upvote: | 228 |
Title: Listening to the audiobook of “The Innovators” (highly recommend) today and the story tells of how Bill Gates wrote Basic for the Altair <i>on an 8080 emulator written by Paul Allen in 1974</i>.<p>How is this feat of computer programming not a major legend in computing history? Surely Paul Allen writing this emulator is even more impressive than Bill writing basic?
Upvote: | 53 |
Title: I’m steadily looking for new opportunities, but am increasingly annoyed by the LinkedIn / Indeed grind. I feel like half the jobs are recruiting firms or very bloated positions with >500 applicants.<p>I love the monthly “Who is hiring?” thread — these positions almost always yield more responses and suffer less from false advertising.<p>Are there other sites I’m not considering? Methods I’m not using? How do you find <i>good</i> (defined as not bloated and optimized for LI) job opportunities in the current market?
Upvote: | 241 |
Title: Hi, I'd like to ask what are the shell scripts you enjoy using or find useful?<p>It might be something you incorporated to your terminal-based workflow. Or maybe some specific scripts that you often reuse. Or you have used it once, but it might be useful to other people. Or maybe you just have a script that is fun to use? Please share<p>My (not anymore) hidden intention is to gather your recommendations to build an open-source shell script registry https://spellbook.maczan.pl/ Source code is here for you if you want to self host or fork it https://github.com/jmaczan/spellbook<p>A script I sometimes use is a commands repeater https://registry.spellbook.maczan.pl/repeat-sh/spell.sh You can specify an interval and a flag to reset/keep the terminal's content after a script invocation<p>Thanks!
Upvote: | 70 |
Title: I wrote a relational database management system (RDBMS) (sqlite clone) from scratch in pure Python.
Upvote: | 371 |
Title: Hey HN! Alan and Albert here, cofounders of Serra. Serra is end-to-end dbt—we make building reliable, scalable ELT/ETL easy by replacing brittle SQL scripts with object-oriented Python. It’s open core: <a href="https://github.com/Serra-Technologies/serra">https://github.com/Serra-Technologies/serra</a>, and our docs are here: <a href="https://docs.serra.io/documentation/">https://docs.serra.io/documentation/</a>.<p>I stumbled into this idea as a data engineer for Disney+’s subscriptions team. We were “firefighters for data,” ready to debug huge pipelines that always crashed and burned. The worst part of my job at Disney+ was the graveyard on-call rotations, where pagers from 12am to 5am were guaranteed, and you'd have to dig through thousands of lines of someone else’s SQL.
SQL is long-winded—1000 lines of SQL can often be summarized with 10 key transforms. We take this SQL and summarize those transforms with reusable, testable, scalable Spark objects.<p>Serra is written in PySpark and modularizes every component of ETL through Spark objects. Similar to dbt, we apply software engineering best practices to data, but we aim to do it not just with transformations, but with data connectors as well.
We accomplish this with a configuration YAML file—the idea is if we have a pipeline with said 1000 line SQL script that is using third-party connectors, we can summarize all of this into a 12-block config file that gives easy high-level overhead and debugging capabilities—10 blocks for the transforms and 2 for the in-house connectors. Then, we can add tests and custom alerts to each of these objects/blocks so that we know where exactly the pipeline breaks and why.<p>We are open-source to make it easy to customize Serra to whatever flavor you like with custom transformers/connectors. The connectors we support OOB are Snowflake, AWS, BigQuery, and Databricks and are adding more based on feedback. The transforms we support include mapping, pivoting, joining, truncating, imputing, and more.
We’re doing our best to make Serra as easy to use as possible. If you have docker installed, you can run this docker command to instantly get setup with a Serra environment to create modular pipelines.<p>We wrap up our functionality with a command line tool that lets you: - create your ETL pipelines, test them locally with a subset of your data, and deploy them to the cloud (currently we only support Databricks, but will soon support others and plan to host our own clusters too). It also has an experimental “translate” feature which is still a bit finicky, but the idea is to take your existing SQL script and get suggestions on how you can chunk up and modularize your job with our config. It’s still just a super early suggestion feature that is definitely not fleshed out, but we think it’s a cool approach.<p>Here’s a quick demo going through retooling a long-winded SQL script to an easily maintainable, scalable ETL job: <a href="https://www.loom.com/share/acc633c0ec03455e9e8837f5c3db3165?sid=341c529c-1c4a-41b2-85c4-ca8589ff965a" rel="nofollow noreferrer">https://www.loom.com/share/acc633c0ec03455e9e8837f5c3db3165?...</a>. (docker command: docker run --mount type=bind,source="$(pwd)",target=/app -it serraio/serra /bin/bash)<p>We don’t see or store any of your data—we’re a transit layer that helps you write ETL jobs that you can send to your warehouse of choice with your actual data. Right now we are helping customers retool their messy data pipelines and plan to monetize by hosting Serra on the cloud, charging if you run the job on our own clusters, and per API call on our translate feature (once it’s mature).<p>We’re super excited to launch this to Hacker News. We’d love to hear what you think. Thanks in advance!
Upvote: | 139 |
Title: I hate talking to people.<p>So I made ChatGPT do it for me.
Upvote: | 69 |
Title: Hi HN<p>I needed a way to monitor network calls made by chrome extensions so I made a small extension.<p>You can install it by dropping the zip or crx into the extensions page. It'll be on the chrome store whenever/if it gets through the review.<p>Hopefully it's useful to others.<p><a href="https://github.com/dnakov/little-rat">https://github.com/dnakov/little-rat</a><p><a href="https://twitter.com/dnak0v" rel="nofollow noreferrer">https://twitter.com/dnak0v</a>
Upvote: | 555 |
Title: Hi!
"Another chatbot for your content" you may ask? Not really! Let me explain.<p>I have realized that there's an untapped potential for ChatGPT plugins: why leave them locked up in the OpenAI store? Let's embed them on websites or share using links!<p>Let's make chatbots ~ do stuff ~, not just talk about what they know.<p>That's how Chatwith was born. It is a chat widget, it scrapes your website, knowledge page, notion etc - but more importantly, it also understands your API. It can interact with that API on your or your visitors' behalf.<p>Some use cases & ideas:<p>- Shopify bot - ask it for order status and invoice<p>- Realtor - ask for budget, provide listings<p>- Survey - ask questions, collect answers, submit to Typeform<p>- Airline service - get your boarding pass<p>- Mixpanel - talk to your analytics<p>In other words:<p>- Talk to APIs & hot data<p>- ChatGPT plugin on every website<p>- Your SaaS can be a chatbot<p>- APIs made accessible (they were reserved for devs until now)<p>- Create an API, use chat as the interface - skip the frontend<p>I am trying to envision a future where website visitors are empowered to just say what they want from the business - instead of navigating endless dashboards and documentation.<p>I'm looking for some feedback about my product! lmk<p>Rafal
Upvote: | 46 |
Title: I've noticed that many of the startups in the YC S23 batch are open source. Is that correct? If so, why has this become a trend?
Upvote: | 42 |
Title: Hey HN! We are building Epsilla (<a href="https://github.com/epsilla-cloud/vectordb">https://github.com/epsilla-cloud/vectordb</a>), an open-source, self-hostable vector database for semantic similarity search that specializes in low query latency.<p>When do we need a vector database? For example, GPT-3.5 has a 16k context window limit. If we want to let it answer a question about a 300 page book, we cannot put the whole book content into the context. We have to choose the sections of the book that are most relevant to the question. Vector database is specialized at ranking and picking the most relevant content from a large pool of documents based on their semantic similarity.<p>Most vector databases utilize hierarchical navigational small world (HNSW) for indexing the vectors for high precision vector search, and its latency significantly degrades when the precision target is higher than 95%.<p>At a previous company, we worked on building the parallel graph traversal engine. We realized that the bottleneck of HNSW performance is because there are too many sequential traversal steps that don't fully leverage multi-core CPU computation resources. After some research, we found that there are algorithms such as SpeedANN that are targeting this problem, which is not leveraged by industry yet. So we built the Epsilla vector database to turn the research into a production system.<p>With Epsilla, we shoot for 10x lower vector search latency compared to HNSW based vector databases. We did an initial benchmark against the top open source vector databases: <a href="https://medium.com/@richard_50832/benchmarking-epsilla-with-some-of-the-top-vector-databases-543e2b7708e5" rel="nofollow noreferrer">https://medium.com/@richard_50832/benchmarking-epsilla-with-...</a><p>We provide a Docker image for you to install Epsilla backend locally, and provide a Python client and a JavaScript client to connect and interact with it.<p>Quickstart:<p><pre><code> docker pull epsilla/vectordb
docker run --pull=always -d -p 8888:8888 epsilla/vectordb
pip install pyepsilla
git clone https://github.com/epsilla-cloud/epsilla-python-client.git
cd examples
python hello_epsilla.py
</code></pre>
We just started a month ago. We'd love to hear what you think, and more importantly, what you wish to see in the future. We are thinking about a serverless vector database on cloud with a consumption based pricing model, and we are eager to get your feedback.
Upvote: | 111 |
Title: I wrote a small Chrome extension that allows the user to replace all usernames on Hacker News with either the string `anonymous` or a deterministic hashed value.
Upvote: | 45 |
Title: Outlines is a Python library that focuses on text generation with large language models. Brandon and I are not LLM experts and started the project a few months ago because we wanted to understand better how the generation process works. Our original background is probabilistic, relational and symbolic programming.<p>Recently we came up with a fast way to generate text that matches a regex (<a href="https://blog.normalcomputing.ai/posts/2023-07-27-regex-guided-generation/regex-guided-generation.html" rel="nofollow noreferrer">https://blog.normalcomputing.ai/posts/2023-07-27-regex-guide...</a>). The basic idea is simple: regular expressions have an equivalent Deterministic-Finite Automaton (DFA) representation. We can transform this DFA into a generative model: in each state we get a list of symbols which correspond to completions that partially match the regular expression. We mask the other symbols in the logits returned by a large language model, sample a new symbol and move to the next state. The subtelty is that language models work with tokens, not symbols, so we derive a new FSM whose alphabet is the model's vocabulary. We can do this in only one pass over the vocabulary.<p>Generating the token masks thus only requires a dictionary lookup at each state. Our method blows other libraries like Microsoft's guidance out of the water.<p>From there it was only a small leap to be able to generate text that follows a JSON schema (<a href="https://json-schema.org/" rel="nofollow noreferrer">https://json-schema.org/</a>), or is parseable into a Pydantic model (<a href="https://docs.pydantic.dev/latest/usage/models/" rel="nofollow noreferrer">https://docs.pydantic.dev/latest/usage/models/</a>). The method works with union types, optional types, nested schemas, arrays, everything. It is guaranteed that the output is parseable.<p>I think it's cool, and I've spent a lot of time watching even tiny models output valid JSON over the weekend. Hope you will too.<p>I look forward to feedback, bug reports, feature requests and discussions!<p>Edit: Link to our pre-print explaining the method and how this can be extended to generate text that follows a Context-Free Grammar <a href="https://arxiv.org/abs/2307.09702" rel="nofollow noreferrer">https://arxiv.org/abs/2307.09702</a>
Upvote: | 854 |
Subsets and Splits