The best Hacker News stories from Show from the past week
Latest posts:
Show HN: Visual intuitive explanations of LLM concepts (LLM University)
Hi HN,<p>We've just published a lot of original, visual, and intuitive explanations of concepts to introduce people to large language models.<p>It's available for free with no sign-up needed and it includes text articles, some video explanations, and code examples/notebooks as well. And we're available to answer your questions in a dedicated Discord channel.<p>You can find it here: https://llm.university/<p>Having written https://jalammar.github.io/illustrated-transformer/, I've been thinking about these topics and how best to communicate them for half a decade. But this project is extra special to me because I got to collaborate on it with two of who I think of as some of the best ML educators out there. Luis Serrano of https://www.youtube.com/@SerranoAcademy and Meor Amer, author of "A Visual Introduction to Deep Learning" https://kdimensions.gumroad.com/l/visualdl<p>We're planning to roll out more content to it (let us know what concepts interest you). But as of now, it has the following structure (With some links for highlighted articles for you to audit):<p>---<p>Module 1: What are Large Language Models<p>- Text Embeddings (https://docs.cohere.com/docs/text-embeddings)<p>- Similarity between words and sentences (https://docs.cohere.com/docs/similarity-between-words-and-sentences)<p>- The attention mechanism<p>- Transformer models (https://docs.cohere.com/docs/transformer-models HN Discussion: https://news.ycombinator.com/item?id=35576918)<p>- Semantic search<p>---<p>Module 2: Text representation<p>- Classification models (https://docs.cohere.com/docs/classification-models)<p>- Classification Evaluation metrics (https://docs.cohere.com/docs/evaluation-metrics)<p>- Classification / Embedding API endpoints<p>- Semantic search<p>- Text clustering<p>- Topic modeling (goes over clustering Ask HN posts https://docs.cohere.com/docs/clustering-hacker-news-posts)<p>- Multilingual semantic search<p>- Multilingual sentiment analysis<p>---<p>Module 3: Text generation<p>- Prompt engineering (https://docs.cohere.com/docs/model-prompting)<p>- Use case ideation<p>- Chaining prompts<p>---<p>A lot of the content originates from common questions we get from users of the LLMs we serve at Cohere. So the focus is more on application of LLMs than theory or training LLMs.<p>Hope you enjoy it, open to all feedback and suggestions!
Show HN: Honda Civic Infotainment Reverse-Engineering
I own a 2021 Honda Civic and have been annoyed by the lack of public documentation/hacking tools for the Android-based headunit. I hope to address this by publishing my research into the headunit and encouraging discussion and community contribution
Show HN: HelpHub – GPT chatbot for any site
Hi HN,<p>I’m the founder of a SaaS platform called CommandBar (YC S20). We’ve been mucking around with AI-related side quests for a while, but recently got excited enough about one to test it with some customers. Results were surprisingly good so we decided to launch it.<p>HelpHub is AI chat + semantic search for any website or web app.<p>You can add source content in 3 ways:
-Crawling any public site via a URL (e.g. your marketing site or blog)
-Syncing with a CMS (like Zendesk or Intercom)
-Add content manually<p>The chatbot is then “trained” on that content and will answer question’s based on that content only, not referencing directly the knowledge.<p>The output is an embeddable widget the contains two things: the chatbot interface for user’s to ask questions, and a search interface for users to search through the content the bot is trained on directly (as well as view source content).<p>You can play around with a demo on some popular sites here: <a href="https://helphub.commandbar.com">https://helphub.commandbar.com</a><p>Some features we added that make it better IMO than just chat:
-Suggested questions (based on the page the user is on and their chat history)
-Suggested follow-up questions in a chat response
-Ask a question about a specific doc
-Recommend content based on who the user is and where they are<p>Would love to hear feedback (not lost on me that there are other chatgpt-for-your-site products and we are probably missing a ton of functionality from there) and can also share details about how we built this. It’s not rocket science but does feel magic :)<p>-James
Show HN: Image background removal without annoying subscriptions
Hi HN,<p>Removing the background from images is a surprisingly common image processing task, and AI has made it really easy. The technology has come a long way since segment leader remove.bg launched here on hn in Dec 2018 [1]. Chasing remove.bg's success, a legion of providers have come on the market offering varying levels of quality & service.<p>Despite there being a large number of competing services, most still price for very high (~95%?) gross margins. Furthermore, subscriptions make the effective unit price a lot higher than the list price for infrequent users, and requires effort & attention to ensure you're getting value for money. This has prevented a host of use cases (e.g. infrequent professional / hobbyist) and business models (e.g. ad-supported websites & mobile apps).<p>We see this as an opportunity where we can jump to the market's logical conclusion to gain market share and build goodwill: cost-plus PAYGO pricing, i.e. the "S3 pricing model".<p>So we've built yet-another image background removal service ( <a href="https://pixian.ai" rel="nofollow">https://pixian.ai</a> - introductory post 6 months ago [2], a ton has been improved since then) but with a couple of twists:<p>1. Quantified quality comparison (90-120% of remove.bg, depending on image category), free for you to check your own images so you can make an informed choice.<p>2. Customer-friendly pricing (PAYGO @ 1-10% of competitors' subscriptions) with a generous free tier (and free while in beta).<p>3. A novel API result format: Delta PNG [3], which offers excellent latency & bandwidth savings. Especially useful for mobile apps.<p>4. Operational transparency: actual volume & latency metrics public, with more coming soon (all API providers should be showing this).<p>There's of course more to it than just price and we see several sources of differentiation in this market: quality, price, capability, reliability, latency, and goodwill.<p>As a new entrant we're looking to meet-or-beat the quality bar; beat on price, capability, reliability and latency; and to build up goodwill over time.<p>Our goal is to make it a no-brainer for new accounts to choose us, and to provide the tools and guidance necessary for existing accounts to make the switch with confidence.<p>We'd love for you to try it out and to hear your thoughts!<p><a href="https://pixian.ai" rel="nofollow">https://pixian.ai</a><p>[1] <a href="https://news.ycombinator.com/item?id=18697601" rel="nofollow">https://news.ycombinator.com/item?id=18697601</a>
[2] <a href="https://news.ycombinator.com/item?id=33439405" rel="nofollow">https://news.ycombinator.com/item?id=33439405</a>
[3] <a href="https://pixian.ai/api/deltaPng" rel="nofollow">https://pixian.ai/api/deltaPng</a>
Show HN: Image background removal without annoying subscriptions
Hi HN,<p>Removing the background from images is a surprisingly common image processing task, and AI has made it really easy. The technology has come a long way since segment leader remove.bg launched here on hn in Dec 2018 [1]. Chasing remove.bg's success, a legion of providers have come on the market offering varying levels of quality & service.<p>Despite there being a large number of competing services, most still price for very high (~95%?) gross margins. Furthermore, subscriptions make the effective unit price a lot higher than the list price for infrequent users, and requires effort & attention to ensure you're getting value for money. This has prevented a host of use cases (e.g. infrequent professional / hobbyist) and business models (e.g. ad-supported websites & mobile apps).<p>We see this as an opportunity where we can jump to the market's logical conclusion to gain market share and build goodwill: cost-plus PAYGO pricing, i.e. the "S3 pricing model".<p>So we've built yet-another image background removal service ( <a href="https://pixian.ai" rel="nofollow">https://pixian.ai</a> - introductory post 6 months ago [2], a ton has been improved since then) but with a couple of twists:<p>1. Quantified quality comparison (90-120% of remove.bg, depending on image category), free for you to check your own images so you can make an informed choice.<p>2. Customer-friendly pricing (PAYGO @ 1-10% of competitors' subscriptions) with a generous free tier (and free while in beta).<p>3. A novel API result format: Delta PNG [3], which offers excellent latency & bandwidth savings. Especially useful for mobile apps.<p>4. Operational transparency: actual volume & latency metrics public, with more coming soon (all API providers should be showing this).<p>There's of course more to it than just price and we see several sources of differentiation in this market: quality, price, capability, reliability, latency, and goodwill.<p>As a new entrant we're looking to meet-or-beat the quality bar; beat on price, capability, reliability and latency; and to build up goodwill over time.<p>Our goal is to make it a no-brainer for new accounts to choose us, and to provide the tools and guidance necessary for existing accounts to make the switch with confidence.<p>We'd love for you to try it out and to hear your thoughts!<p><a href="https://pixian.ai" rel="nofollow">https://pixian.ai</a><p>[1] <a href="https://news.ycombinator.com/item?id=18697601" rel="nofollow">https://news.ycombinator.com/item?id=18697601</a>
[2] <a href="https://news.ycombinator.com/item?id=33439405" rel="nofollow">https://news.ycombinator.com/item?id=33439405</a>
[3] <a href="https://pixian.ai/api/deltaPng" rel="nofollow">https://pixian.ai/api/deltaPng</a>
Show HN: Image background removal without annoying subscriptions
Hi HN,<p>Removing the background from images is a surprisingly common image processing task, and AI has made it really easy. The technology has come a long way since segment leader remove.bg launched here on hn in Dec 2018 [1]. Chasing remove.bg's success, a legion of providers have come on the market offering varying levels of quality & service.<p>Despite there being a large number of competing services, most still price for very high (~95%?) gross margins. Furthermore, subscriptions make the effective unit price a lot higher than the list price for infrequent users, and requires effort & attention to ensure you're getting value for money. This has prevented a host of use cases (e.g. infrequent professional / hobbyist) and business models (e.g. ad-supported websites & mobile apps).<p>We see this as an opportunity where we can jump to the market's logical conclusion to gain market share and build goodwill: cost-plus PAYGO pricing, i.e. the "S3 pricing model".<p>So we've built yet-another image background removal service ( <a href="https://pixian.ai" rel="nofollow">https://pixian.ai</a> - introductory post 6 months ago [2], a ton has been improved since then) but with a couple of twists:<p>1. Quantified quality comparison (90-120% of remove.bg, depending on image category), free for you to check your own images so you can make an informed choice.<p>2. Customer-friendly pricing (PAYGO @ 1-10% of competitors' subscriptions) with a generous free tier (and free while in beta).<p>3. A novel API result format: Delta PNG [3], which offers excellent latency & bandwidth savings. Especially useful for mobile apps.<p>4. Operational transparency: actual volume & latency metrics public, with more coming soon (all API providers should be showing this).<p>There's of course more to it than just price and we see several sources of differentiation in this market: quality, price, capability, reliability, latency, and goodwill.<p>As a new entrant we're looking to meet-or-beat the quality bar; beat on price, capability, reliability and latency; and to build up goodwill over time.<p>Our goal is to make it a no-brainer for new accounts to choose us, and to provide the tools and guidance necessary for existing accounts to make the switch with confidence.<p>We'd love for you to try it out and to hear your thoughts!<p><a href="https://pixian.ai" rel="nofollow">https://pixian.ai</a><p>[1] <a href="https://news.ycombinator.com/item?id=18697601" rel="nofollow">https://news.ycombinator.com/item?id=18697601</a>
[2] <a href="https://news.ycombinator.com/item?id=33439405" rel="nofollow">https://news.ycombinator.com/item?id=33439405</a>
[3] <a href="https://pixian.ai/api/deltaPng" rel="nofollow">https://pixian.ai/api/deltaPng</a>
Show HN: Open sourcing Harmonic, my Android Hacker News client
Show HN: SpaceBadgers – Free and Libre SVG Badges
Greetings, Hacker News community!<p>I am thrilled to present SpaceBadgers, a new free and open-source SVG badge generator I've been working on. It's located at badgers.space.<p>SpaceBadgers is born out of the desire to offer more flexibility and customization for project badges, often used in open-source projects.<p>It's fully open source, provided under the permissive MIT license, and will always be provided for free. The core badge worker is written in Rust, and so is the library behind it, which you can also find on crates.io under the name spacebadgers.<p>I am excited to receive your feedback and suggestions. Check it out and let me know what you think in the comments. Contributions are also welcomed and appreciated. You can find the source code here: <a href="https://github.com/splittydev/spacebadgers">https://github.com/splittydev/spacebadgers</a>.
Show HN: Trogon – An automatic TUI for command line apps
Hi HN,<p>Trogon is a project to generate a TUI for command line apps.<p>It presents the arguments, options, and switches as a form. Editing the form generates a command line, which you can then run with a keypress.<p>I'm a lover of the command line. But I can recall only a fraction of the switches for most commands I use. I would love it if there was a TUI available for most commands.<p>Trogon currently works with Python and the Click library, but I would like it to cover more of the Python ecosystem and also generate TUIs for apps not written in Python.<p>More information in the repository.<p>Let me know what you think...
Show HN: YouTube Full Text Search – Search all of a channel from the commandline
yt-fts is a simple python script that uses yt-dlp to scrape all of a youtube channels subtitles and load them into an sqlite database that is searchable from the command line. It allows you to query a channel for specific key word or phrase and will generate time stamped youtube urls to the video containing the keyword.
Show HN: Willow – Open-source privacy-focused voice assistant hardware
As the Home Assistant project says, it's the year of voice!<p>I love Home Assistant and I've always thought the ESP BOX[0] hardware is cool. I finally got around to starting a project to use the ESP BOX hardware with Home Assistant and other platforms. Why?<p>- It's actually "Alexa/Echo competitive". Wake word detection, voice activity detection, echo cancellation, automatic gain control, and high quality audio for $50 means with Willow and the support of Home Assistant there are no compromises on looks, quality, accuracy, speed, and cost.<p>- It's cheap. With a touch LCD display, dual microphones, speaker, enclosure, buttons, etc it can be bought today for $50 all-in.<p>- It's ready to go. Take it out of the box, flash with Willow, put it somewhere.<p>- It's not creepy. Voice is either sent to a self-hosted inference server or commands are recognized locally on the ESP BOX.<p>- It doesn't hassle or try to sell you. If I hear "Did you know?" one more time from Alexa I think I'm going to lose it.<p>- It's open source.<p>- It's capable. This is the first "release" of Willow and I don't think we've even begun scratching the surface of what the hardware and software components are capable of.<p>- It can integrate with anything. Simple on the wire format - speech output text is sent via HTTP POST to whatever URI you configure. Send it anywhere, and do anything!<p>- It still does cool maker stuff. With 16 GPIOs exposed on the back of the enclosure there are all kinds of interesting possibilities.<p>This is the first (and VERY early) release but we're really interested to hear what HN thinks!<p>[0] - <a href="https://github.com/espressif/esp-box">https://github.com/espressif/esp-box</a>
Show HN: I built my first Cyberdeck
Show HN: Moderator Mayhem, a game about the difficulties of content moderation
Show HN: Boring Report, a news app that uses AI to desensationalize the news
In today's world, catchy headlines and articles often distract readers from the facts and relevant information. By utilizing OpenAI's language models, Boring Report processes sensationalist news articles, transforms them into the content you see, and helps readers focus on the essential details. We recently updated our iOS app experience, so any and all feedback would be appreciated.<p>App Link: <a href="https://apps.apple.com/us/app/boring-report-news-by-ai/id6446786839?itsct=apps_box_link&itscg=30200" rel="nofollow">https://apps.apple.com/us/app/boring-report-news-by-ai/id644...</a>
Show HN: Build progressively enhanced reactive HTML apps using Go and Alpine.js
Fir leverages Golang’s standard library html/template package and a bit of alpinejs to allow building reactive UIs. You start with plain old html and use alpinejs to enhance it to bring no-page-reload interactivity to web apps.<p>The Fir toolkit is designed for Go developers with moderate html/css & js skills who want to progressively build reactive web apps without mastering complex web frameworks. It includes a Go library and an Alpine.js plugin.<p>How it works ?<p>On receiving user-interactions the fir server re-renders html templates and sends it over the wire where the fir client library selectively updates the changed areas.<p>When a user event is received by a Fir route, an array of html templates are rendered on the server and returned as an array of DOM events to the browser. The DOM events are consumed by the alpinejs plugin and dispatched within the DOM where listeners attached to elements can use the event to update the DOM.<p>See the demo and quickstart here: <a href="https://livefir.fly.dev/" rel="nofollow">https://livefir.fly.dev/</a>
Show HN: Engineering Book Club
An online community that enjoys reading and discussing engineering books together.
Show HN: Cptn.io – open-source integration platform
Hi, I am Krishna Thota. I am building an open source integration and data platform(<a href="https://cptn.io" rel="nofollow">https://cptn.io</a>). The product is MIT licensed and the repo is at <a href="https://github.com/cptn-io/el-cptn">https://github.com/cptn-io/el-cptn</a>.<p>I have started on my startup journey an year ago and launched a monitoring platform called DevRaven. Unfortunately the product did not takeoff as expected. That story is for another day.<p>But during the course of building the product, I have built several integrations leveraging MQs and Cloud Functions. While building and deploying Cloud Functions for happy paths is easy, I had to monitor logs for failures, build retry mechanisms or manually process failed events, keep instances running to prevent cold start timeouts. It can also get expensive with charges for MQs, server time for running cloud functions etc and costs can be unpredictable.<p>I thought of building a platform where I can build integrations quickly, have the ability to look at incoming/outgoing events, look at logs, retry any failed events etc. And finally, predictable costs for running the infrastructure. cptn.io provides all these capabilities and more. You can build pipelines to integrate with any cloud services, send data from your backend to data warehouses, listen to web hook events etc. The platform can be integrated into any stack by sending events to HTTP end points.<p>Instead of trying to build a business first or launch an open source product under restrictive licenses, the platform will be available under MIT license so any user or customer can use it. There is no ee folder or complex dual licensing and I am also committing to releasing SSO under MIT. The plan is to offer a managed service in the cloud at a later time, accept sponsors for prioritizing features for enterprise customers and charge for enterprise support.<p>It should take less than 5 minutes to get the platform running on your machine. Welcome any feedback, feature requests, PRs and bug reports.
Show HN: SineRider, a math puzzle game
Messing with your TI-84 graphing calculator is a rite of passage for every teenager who has ever been bored in a math class. In 2013 I was that teenager, and it gave me an idea for a tiny game about sledding on graphs. This project grew into my white whale, and I spent my twenties trying and failing to finish it alone. I shelved the game when I started working for Hack Club in 2018—until last May, when a few community members took it off the shelf. The project took on a life of its own, and turned into a year of nights and weekends from a global team of 20+ teens in 8+ countries. Today SineRider enters public beta!<p>SineRider is literally an infinite universe of function composition puzzles, each with infinite solutions, that range from welcoming for 9th graders to difficult for even the most serious matlab user. And every day we tweet out a fresh one to be solved with your morning coffee.<p>We hope you enjoy playing SineRider as much as we’ve enjoyed making it. And we’re not done! Mobile support, polar coordinates, and a level editor are all on the roadmap. SineRider is a living project, to be continuously built and maintained as free OSS by the Hack Club community: <a href="https://github.com/hackclub/sinerider">https://github.com/hackclub/sinerider</a><p>The team that built the game will try to be in the comments today between high school classes and AP tests.<p>—chris walker, creative director<p>Watch the trailer: <a href="https://www.youtube.com/watch?v=35nDYoIwiA8">https://www.youtube.com/watch?v=35nDYoIwiA8</a><p>Play now: <a href="https://sinerider.com" rel="nofollow">https://sinerider.com</a>
Show HN: Git Hooting
00's called, they want their RSS feeds back.<p>I was looking at my growing Github gist collection when a sudden urge to blog and make a name for myself "by not programming" struck. Part way into implementing my oh so special static website generator it occurred to me that, quite frankly, Github gists is a pretty decent publishing platform. I mean, it gives you reasonably extended markdown with previews, heck I could even write in org-mode, has comments, follower - followee relationship, extended search with filters, check out locally and push your edits. Did someone say "edit button"?<p>Thus the idea behind <a href="https://git.ht" rel="nofollow">https://git.ht</a> was born: collect gists into RSS feeds and force everyone, kicking and screaming, into the good old days when Google Reader was king. Well, it's a bit more than that now. But basically, you create a gist or grab an old one, name its main file `hoot.md` or `hoot.org` if org-mode is your poison, make it public and voila. These "hoots" make it into your RSS feed and will get permalinks with social graph metatags, so you get nice previews when you share them on Twitter and such.<p>To take it for a spin:
- pick a subdomain e.g. foo.git.ht,
- navigate you browser there,
- login with Github.<p>I still consider it alpha, but it should work. Report any issues as you would normally on Github <a href="https://github.com/fullmeta-dev/githoot-public">https://github.com/fullmeta-dev/githoot-public</a>.<p>Thank you
Show HN: Git Hooting
00's called, they want their RSS feeds back.<p>I was looking at my growing Github gist collection when a sudden urge to blog and make a name for myself "by not programming" struck. Part way into implementing my oh so special static website generator it occurred to me that, quite frankly, Github gists is a pretty decent publishing platform. I mean, it gives you reasonably extended markdown with previews, heck I could even write in org-mode, has comments, follower - followee relationship, extended search with filters, check out locally and push your edits. Did someone say "edit button"?<p>Thus the idea behind <a href="https://git.ht" rel="nofollow">https://git.ht</a> was born: collect gists into RSS feeds and force everyone, kicking and screaming, into the good old days when Google Reader was king. Well, it's a bit more than that now. But basically, you create a gist or grab an old one, name its main file `hoot.md` or `hoot.org` if org-mode is your poison, make it public and voila. These "hoots" make it into your RSS feed and will get permalinks with social graph metatags, so you get nice previews when you share them on Twitter and such.<p>To take it for a spin:
- pick a subdomain e.g. foo.git.ht,
- navigate you browser there,
- login with Github.<p>I still consider it alpha, but it should work. Report any issues as you would normally on Github <a href="https://github.com/fullmeta-dev/githoot-public">https://github.com/fullmeta-dev/githoot-public</a>.<p>Thank you