Show HN: Dobb·E – towards home robots with an open-source platform Hi HN! Proud to share our open-source robot platform, Dobb·E, a home robot system that needs just 5 minutes of human teaching to learn new tasks. We've already taken Dobb·E to 10 different homes in New York, taught it 100+ tasks, and we are just getting started! I would love to hear your thoughts about this. Here are some more details, below (or see a Twitter thread with attached media: https://twitter.com/i/status/1729515379892826211 or https://ift.tt/h1GbNlM ): We engineered Dobb·E to maximize efficiency, safety, and user comfort. As a system, it is composed of four parts: a data collection tool, a home dataset, a pretrained vision model, and a policy fine-tuning recipe. We teach our robots with imitation learning, and for data collection, we created the “Stick”, a tool made out of $25 of hardware and an iPhone. Then, using the Stick, we collected a 13 hour dataset in 22 New York homes, called Homes of New York (HoNY). HoNY has 1.5M frames collected over 216 different "environments" which is an order of magnitude larger compared to similar open source datasets. Then we trained a foundational vision model that we can fine-tune fast (15 minutes!) on a new task with only 5 minutes (human time)/ 90 seconds (demo time) of data. So from start to finish, it takes about 20 minutes to teach the robot a new task. Over a month, we visited 10 homes, tried 109 tasks, and got 81% success rate in simple household tasks. We also found a line of challenges, from mirrors to heavy objects, that we must overcome if we are to get a general purpose home robot. We open-sourced our entire system because our primary goal is to get more robotics and AI researchers, engineers, and enthusiasts to go beyond constrained lab environments and start getting into homes! So here is how you can get started: 1. Code and STL files: https://ift.tt/vLDYskV 2. Technical documentation: https://ift.tt/e0Vkrbu 3. Paper: https://ift.tt/cSeQHTh 4. More videos and the dataset: https://dobb-e.com 5. Robot we used: https://hello-robot.com https://dobb-e.com/ November 29, 2023 at 04:53AM
Show HN: Neets.ai – Cheap and Robust TTS Platform We used to rely on ElevenLabs for their TTS API, however the prices for Eleven and their competitors were enormous. That led us to developing our own TTS platform, Neets.ai, which we will grow and improve with a much lower price than our peers. We're always looking for developers. If you think you have the skills, reach out: https://ift.tt/RxFygYX https://neets.ai/ November 29, 2023 at 03:30AM
Show HN: Build an open-source computer vision model in seconds using text Hello HN! I want to share something me and a few friends have been working on for a while now — Zeroshot, a web tool that builds image classifiers using text-image models and autolabeling. What does this mean in practice? You can put together an image classifier in about 30 seconds that’s faster and more accurate than CLIP, but that you can deploy yourself however you’d like. It’s open source, commercially licensed, and doesn’t require you to pay anyone per API call. Here's a 2 minute video that shows it off: https://www.youtube.com/watch?v=S4R1gtmM-Lo How/why does it work? We believe that with the rise of foundation vision models, computer vision will fundamentally change. These powerful models will let any devs “compile” a model ahead of time with a subset of the foundation model’s characteristics, using only text and a web-tool. The days of teams of MLEs building complex models and pipelines are ending. Zeroshot works by using two powerful pre-trained models, CLIP and DINOv2 together. The web-app allows users to quickly create our training sets via text search. Using pre-cached DINOv2 features, we generate a simple linear model that can be trained and deployed without any fine-tuning. Since you can see what’s going into your training set, you can tune your prompts to get the type of performance or detail you want. CLIP Small -- Size: 335 MB, Latency: 35ms CLIP Large -- Size: 891 MB, Latency: 276ms Zeroshot -- Size: 85 MB, Latency: 20ms What’s next? We wanna see how people use or would use the tool before deciding what to do next. On the list: clients for iOS and NodeJS, speeding up GPU inference times via TensorRT, offering larger Zeroshot models for better accuracy, easier results refining, support for bringing your own data lake, model refinement using GPT-V, we’ve got plenty of ideas. https://ift.tt/lGidfT1 November 29, 2023 at 01:48AM
Show HN: I built a domain name marketplace for folks (like me) who hoard domains Problem: I (accidentally) hoard domains :') 1. I get excited about a new project 2. I buy a domain 3. I get busy, and the domain collects a thick layer of dust I know I'm not alone in this, either So, I had the idea of creating a simple and casual marketplace for folks like me to list their domains at a fair price with a nice community feel to free up these caged domains It felt like a great project for me to pick up some new skills, so I got to it All up, it took me about a month, and I built the whole thing live on Twitch I've always sat on the design, marketing and front-end side of the fence, so this was my first attempt at making a 'full' web app Here's the stack I used: - SvelteKit ( https://kit.svelte.dev/ ) - Supabase ( https://supabase.com/ ) - Resend ( https://resend.com/ ) - ShadCN Svelte ( https://ift.tt/zIUHaAN ) It was super fun to build, and as a beginner, I learnt so much I leaned on AI quite heavily to help advance my speed of groking certain concepts within both SvelteKit & Supabase, and I blogged about the experience and my learnings here: https://ift.tt/z4RnTx1 Any feedback/criticism very much welcome, I've got a lot to learn :) https://ift.tt/eQEtAPO November 27, 2023 at 03:56AM
Show HN: Building an Investment Management Powerhouse The idea for Stockle came from simply having no usable tools to track, manage and explore investments. It felt like every single one existing platforms lacked even the most basic functionalities. In addition the UIs were not focusing on user experience. The lack of detail in the existing platforms led us to try out a different approach to investment management. Focus points for Stockle have been Modern UI, Detailed coverage on investments and flexible portfolio creation. For future we are focusing on even more personalised features. Feel free to put Stockle-beta to test whether you are an investor or not! Many features still under development but core features are there! Maybe it can even spark up your inner Warren Buffet. Feedback is also appreciated! https://stockle.app/ November 25, 2023 at 10:36PM
Show HN: I saw this mind-blowing experiment, so I made a simple version of it Two browser windows (acting as socket clients) communicate their: - Screen dimensions - (screen.width, screen.height) - Window dimensions - (window.innerWidth, window.innerHeight) - Window X/Y position - (window.screenX, window.screenY) ...or whichever calculation works best for you. The original work by Bjorn Staal https://twitter.com/_nonfigurativ_/status/172732259457002734 used localStorage, but I found sockets more fun, because if tweaked a bit, this can be shared with friends. Here's a demo of how it works and the codebase: https://ift.tt/z4pbcyI https://ift.tt/LZlf4HE November 25, 2023 at 07:46PM
Show HN: I Wrote a Book on PostgreSQL for Rails Hi HN. As a backend developer using PostgreSQL and Ruby on Rails for more than 10 years, around 3 years ago, I caught a spark to learn as much as I could about how to operate high performance, scalable, and reliable PostgreSQL databases. As I reflected on my own career path, I realized there are other developers that may want to learn more about this. I saw an opportunity to help show backend developers how they could put advanced, powerful, PostgreSQL features to use, to help build and scale their Rails apps. I've been fortunate to give presentations in both the PostgreSQL and Rails communities over the last couple of years. In that time I've continued to work with these technologies every day. With that said, I'd love for you to check out my new book: "High Performance PostgreSQL for Rails". Currently in ebook/PDF Beta form. While it's only an ebook for now, there's an expected physical version coming in early 2024. Today you can pick up the PDF/ebook version for 40% off using code: turkeysale2023 There are free PDF excerpts you can check out, to see if this book is for you, or something you'd recommend to others. The book is currently on the weekly Best Seller list at Pragmatic Programmers, where it's appeared several times in the last couple of months. The book focuses exclusively on open source software like PostgreSQL and Ruby on Rails, adding more than 40 Ruby gems and PostgreSQL extensions into the stack of tools. The Rails application used throughout the book is available on GitHub. Whether you work with Rails or another similar MVC full-stack framework or ORM like Laravel, Django, or Prisma, I hope there’s something useful in the book for you or your colleagues. I love writing, learning, and sharing, and advocating for PostgreSQL and Ruby on Rails! Thanks for your support! https://ift.tt/xEA4Sbr November 25, 2023 at 01:29AM
Show HN: UI Library Creator Hey HN! The "Generative" trend is booming, and UI Library Creator is our original approach to it. In the last three years, we have added 60+ professional UI libraries to Shuffle's catalog (Shuffle = visual editor for web developers). Still, we know we need more than this to satisfy our growing user base! That's why we created the UI Library Creator. In this tool, you can combine elements and styles to create unique UI libraries that work seamlessly with the Shuffle Editor and all its capabilities (drag-and-drop, customizations, live preview, and more). We provide you with UX solutions (components) written in Tailwind CSS and presets so you can quickly combine them to create what you need. You don't need to talk to a "black box" AI with a chat interface. Possible combinations are in gazillions. We aim for original creations, but you have complete control over the final effect. How to use the UI Library Creator: * Visit: https://ift.tt/2euSfXj * We recommend starting by selecting Assets and Copywriting for your target audience. * When these two options are locked, use the "Shuffle Styles" button to bootstrap your project with the first style. * If you like something, lock the category and then repeat shuffling. You can also change options manually, but with "Shuffle Styles," you can quickly see many creations. If you enjoy the final result, click "Publish now" and send your UI Library to Shuffle. Once processed, it will be available for use in your Dashboard. Let us know what you think! Video (2min) with product tour: https://www.youtube.com/watch?v=8ZFlWEDr7XM https://ift.tt/2euSfXj November 22, 2023 at 07:21PM
Show HN: Cyberkitty – an experimental hybrid-GPT+Human long-form fiction editor NOTE: If you're one of the first people to see this, you can also access a running installation at https://cyberkitty.ai It'll run until I hit my GPT-4 budget limit, then stop working. Be warned that it's in single user, so nothing is private! ---- Last week I came down with a fever and decided to run some experiments on whether the new GPT could generate long-form fiction. Around the same time, I saw a HN comment about how technology gains have lead SWEs to work at higher levels of abstraction—and I thought “huh, why not writers?” Anyway so I was locked indoors for a week and a half and I spent it coding an answer that question. Most of the magic sauce here is just in making it feel really snappy and good to edit—making it easy for the human to collaborate with the objectively faster LLM. The underlying prompt engineering is cool, but it’s the feeling that you can do more than just shoot a prompt at ChatGPT and hope for a good result, or start writing and hope for a good autocomplete. Definitely gave me some ideas for future LLM experiences I want to try out, let me know what you think! https://ift.tt/QPcNtoB November 23, 2023 at 03:41AM
Show HN: GJP-4-GPT, a Gradual JSON Parser for GPT's Often it's easiest to have an LLM output the results you need in JSON form, but streaming those outputs into an actual viewport is nontrivial. Thus I made this package, a streaming JSON parser that handles all the work of decoding your JSON so you can focus on displaying your UIs. https://ift.tt/KTQheHq November 23, 2023 at 02:23AM
Show HN: Plantarium – Create 3D Plant Models with Blender-Style Nodes Plantarium is my open source side project that allows you to created procedural 3D plants in your browser using blender style nodes. Recently I added a public gallery so you can share your plant designs and download others. https://ift.tt/pRzZw6P https://ift.tt/MYpJRPj November 22, 2023 at 05:32PM
Show HN: Gaming Gaiden: A lightweight gaming time tracker in PowerShell Hi Everyone, I made a Powershell application to track time spent playing pc / emulated games. It provides many other interesting statistics as well over long term that can add value to someone's gaming hobby imo. I used to use a closed source program called Gameplay Time Tracker for time tracking focused on gaming, but overtime it started failing and throwing exceptions everywhere. I tried reaching to its Developer to ask if they can open source it and i can fix the problems. I couldn't raise them. It had been truly abandoned. After a while i just decided to build a replacement. Few choices that may seem strange, using Powershell for a GUI application, building the GUI with HTML, CSS and JS instead of native windows forms. All of these were made because i am not a windows app dev / GUI dev. I am infra engineer and have strong scripting skills. Instead of getting bogged down in learning a whole programming language like C# and .Net., decided to go with Powershell. DataTables and ChartJs made it trivial to create the UI. Pragmatic choices over conventional wisdom basically. It turned out well in my opinion. Got some attention on reddit and in a youtube channel. Thought it would be good to share here as well. https://ift.tt/FkVS6mi November 21, 2023 at 11:10PM
Show HN: I've Published 28B Molecule Embeddings on AWS Open Data I’ve finally finished a project that involved gathering 7 billion small molecules, each represented in SMILES notation and having fewer than 50 “heavy” non-hydrogen atoms. Those molecules were “fingerprinted”, producing 28 billion structural embeddings, using MACCS, PubChem, ECFP4, and FCFP4 techniques. These embeddings were indexed using Unum’s open-source tool USearch, to accelerate molecule search. This extensive dataset is now made available globally for free, thanks to AWS Open Data. You can find the complete data sheet and scripts for data visualization on GitHub. https://ift.tt/ljD0U3d November 21, 2023 at 10:00PM
Show HN: I made a conversational language learning chat UI with gpt4-1106 Hi, I built JoJo, a personal conversational language learning chat interface. video demo here: https://ift.tt/1rE0ZQh You can chat casually in both your native and foreign language, translate from your native to foreign language, and get grammar explanations. There is explanation mode when having a conversation which corrects and explains what you type. This can be turned off so you can just chat back and forth in your foreign language. I have a 1395 day streak with duolingo Italian so I wanted to build something to help with conversational practice. There are 15 languages supported to choose from. Any feedback or suggestions is greatly appreciated. Thank you - Kyle https://www.jojo.so/ November 20, 2023 at 08:27PM
Show HN: Serverless VPN App for Android. Unlimited Devices. Unlimited Data Hello HN, Serverless VPN app for Android is fully open source including desktop apps: https://ift.tt/Oknm3R5 Get it on Play Store: https://ift.tt/RxCoIfJ... Feedback about product, pricing, website is most welcome! Thank you, Himanshu https://ift.tt/we38CD1 November 20, 2023 at 12:52AM
Show HN: cmaps.io – Transform Thoughts into Visual Masterpieces Hey! Just launched my new project! It's a tool to rethink the way we study and take notes. Human thinking is inherently associative, not linear. Check it out at https://cmaps.io I would love to get some feedback! Thanks! https://www.cmaps.io/ November 20, 2023 at 02:25AM
Show HN: YouTube banned adblockers so I built an extension to skip their ads Hi HN! Since Youtube no longer allows AdBlockers, I built my own extension to get around their video ads. If there is an ad it temporarily manipulates the video; Mutes the volume, sets speed to 10x, and skips it if there is a button. Chrome Webstore link: https://ift.tt/8Fcrd9E Code: https://ift.tt/d4TjKkL November 19, 2023 at 06:24AM
Show HN: Bemi – data versioning and time travel for PostgreSQL Hi HN, we’re Evgeny, Arjun, and Donna - the builders of Bemi ( https://bemi.io ). Bemi is an enhancement of your existing transactional database to unlock time travel querying. We enable engineers to track all PostgreSQL state changes. We used to all work together at a startup five years ago, and now have got the gang back together to work on Bemi! Many open-source libraries already track data changes, but they’re not reliable since they don’t track changes made outside ORMs with direct SQL queries. Additionally, they come with runtime performance overhead since they’re making database inserts in callbacks. We’ve also consistently seen hacks that engineers write on top of the libraries to store additional context or make it easier for them to query. There are databases already available for event sourcing, but we take a fundamentally different approach. We don’t want engineers to have to rearchitect their existing code, switch to highly specialized databases, or use unnecessary git-like abstractions. We want your system to work the way it already does with your existing database to allow keeping things as simple as possible. Bemi uses change data capture (CDC) by connecting to a built-in replication log of a database, ingesting all changes on the database level, and storing them in a structured format in a destination database. The database connection details can be securely configured through our dashboard UI in a few seconds. People also try adding database triggers to track low-level data change. The main difference is that Bemi integrates not just on the database level, but also on the application level with packages in the most popular frameworks (being released soon) and a Bemi worker automatically stitches application metadata with the lower-level change data to get the “where” (API endpoint, worker, and so on) and “who” (user, cron job, and so on) behind a change. This allows being able to revert all data changes made within an API request, for example. For the sake of transparency, we intend to make money by implementing usage-based pricing and a multi-cloud hosted option. Bemi’s free for early adopters. We are still early though and would love your feedback. What do you think HN? https://bemi.io/ November 17, 2023 at 09:42PM
Show HN: Challenge Your AI Agents/custom GPTs – Can They Survive My Tester? I've pivoted to create AI Agent Tester, a SaaS platform dedicated to testing and validating AI agents from OpenAI, Character.ai, and FlowGPT.com. The platform offers: - Hassle-Free Setup: Get your first simulation up and running in just 5 minutes. - Realistic Simulations: Rigorous user interaction tests to prep your AI agent for real-world scenarios. - Intelligent Scoring: Each response is meticulously evaluated, scoring from 0 to 100(GPT-4 smart enough to evaluate it for you). - Broad Platform Support: Initially rolling out with OpenAI, Character.ai, and FlowGPT.com compatibility. The link: https://ift.tt/tmkvVg3 Eager to hear your thoughts and discuss improvements! https://ift.tt/tmkvVg3 November 17, 2023 at 07:46PM
Show HN: Beta test Execute Program's interactive "Python for Programmers" course I'm Gary Bernhardt, founder of Execute Program. Our "Python for Programmers" course is in a free open beta for the next week or so. We don't normally do open betas, but the infrastructure behind this course is new and very complex, so we want to stress test it. https://ift.tt/aboRVwv... Today, "Python for Programmers" contains 581 interactive code examples covering the core language. It's aimed at established programmers, not beginners. We don't explain basic language features like `while`, but we do show them briefly and note anything special about how they work in Python. We pay special attention to foot guns. For example, we have an entire lesson about Python's mutable default argument foot gun. This is the first of two courses, with the second coming in 2024. For this course, we drew the line at __dunder__ methods: if a topic requires a dunder method other than `__init__`, then it'll be in the follow-up course. This beta is concurrent with the tail end of our editing process, so you may see the course grow by another 17 lessons (214 code examples) during the beta. Some details about how the course works internally, and why we need a beta at all: First, all Python code in the course runs in your browser via Pyodide. (Reality continues to look more and more like my PyCon 2014 talk [1].) You'll feel a pause when the first code example runs, as your browser loads and boots CPython (around 12 MB). After that, it should be as responsive as a local app. Second, if you look at the course page, you'll see that it's structured as a DAG, similar to a "tech tree" in Civilization, Age of Empires, Stellaris, Satisfactory, etc. (Some of those games have true trees, but some of their "trees" are actually DAGs like ours.) You make progress through the course by traversing one graph edge at a time. Our courses have always been structured as graphs internally, but the raw graphs are simply unreadable due to the number of edges [2]. This year, I taught Execute Program to simplify its own course graphs by breaking them into the level subgraphs that you see on the page, so we can finally render them. It automatically turns the mess that I linked above into the clean graphs that you see in the course. The graph for this course is currently a bit dull, but it'll fill out as we finish editing the remaining lessons. I like Everyday TypeScript's graph [3] the best. Please try the course and use the "Give Feedback" entry in the menu to tell us what you think! I'll also stick around in this thread today. [1] https://ift.tt/0Gr8MR2... [2] https://ift.tt/nbJA89W... [3] https://ift.tt/7iFEXUy https://ift.tt/dOYSlvB November 17, 2023 at 02:51AM
Show HN: Version code, models, & datasets together in GitHub Hi HN! We just launched a GitHub integration that scales your Git repos to handle 100 terabytes of files in a single repo. XetData enables data scientists and machine learning engineers to version code, models, and datasets together. Most teams have glued together clunky workflows using S3, DVC, Git, Git LFS, and other tools and make true reproducibility difficult: https://ift.tt/qb7vApR We instead embrace and extend Git so end-users don’t need to learn a new tool and a new set of commands. Our implementation is similar to Git LFS, where we take over the .gitattributes file, push pointers to large files in GitHub, and push the raw, large files to us. We have a few distinct features that we’re proud of that improve the user experience: - Our XetData bot comments on your pull requests to provide links to useful dataset views and model diffs. We’re working on rendering these inside GitHub itself using browser extensions. - Git LFS and similar tools only implement file-level deduplication. We created a new technique called block-based deduplication (published in CIDR’23 conference) specifically for data and ML workflows. The ML lifecycle consists of making lots of iterative changes and our technique helps save storage and time spent downloading and uploading changes. - You can mount large repos to your local machine using git-xet mount for exploratory work. Individual files that are needed are streamed in just in time behind the scenes. We open sourced our implementation of mount and it was well received here on HN: https://ift.tt/tgiGnYo - To give more users access to your data, just add them to your GitHub repo. This is a beta product and we would love all of your feedback. You can find all instructions to try this out here: https://ift.tt/WICGak5 While we’re in beta, our product is completely free to use. We have a Slack you can join or a GitHub issue tracker. - Slack: https://ift.tt/CKYwDBh - GitHub: https://ift.tt/nq3uCNg November 16, 2023 at 11:56PM
Show HN: AI Receptionist, Speaks 64 Languages If you are looking for a receptionist for your startup, consider using Lomni. Lomni is an AI receptionist that can: * Read your website to answer questions * Upsell your product or service according to your instructions * Send texts and emails to you or the caller, for example: * Send links and confirmation texts to callers * Forward customer feedback to support@startup.com * Connect to any API or webhook, in order to: * Generate leads * Book appointments * Process payments * Integrate with any application * Read from and write to your database * Do all of the above in any language, all day, everyday. * Do all of the above over SMS messaging or a chatbot that you could embed into your website. * Coming very soon: Do all of the above over WhatsApp, Facebook Messenger, and Google’s Business Messages. * Coming very soon: generate actionable insights from conversational data using state-of-the-art machine learning. Pricing: Lomni can do all of the above starting at $89/month, approximately 9 cents per minute of calling. This is 1/10th to 1/20th of the standard pricing of calling centres. Custom Demo: Set up in 30 minutes using our user-friendly dashboard and start testing. Alternatively, email hello@lomni.ai for a free custom demo. https://lomni.ai/ November 16, 2023 at 01:59AM
Show HN: Multi-Object Tracking in Python Hello! I've created a small library for tracking, along with a tutorial. I plan to continue developing it. Tracking is an important topic, closely related to object detection. However, I've noticed that it doesn't receive as much attention compared to machine learning approaches. Or, the focus is on filters like the Kalman filter. This tutorial begins with single object tracking and progressively complicates the tasks, introducing various models and a hypothesis tree to solve them. https://ift.tt/tfOjPm7 November 16, 2023 at 12:06AM
Show HN: Movis – A Video Editing Library in Python Originally, this library was created due to my motivation to make videos in Python. Of course, I considered other Python libraries, but I decided to create a new library for video editing because I needed the following features and also for study purposes: - The existence of composition. Especially, the ability to perform advanced video editing by inserting compositions within compositions. - The ability to add animations based on keyframes. - Effects like chroma key and drop shadow. - A caching mechanism to render the same frame quickly. - Various blending modes. These features, despite being standard in proprietary video editing software, were not present in previous Python libraries. Currently, I am using this library to automatically generate videos from raw materials. I would be grateful for any comments, thoughts, or requests. https://ift.tt/aG2J73e November 15, 2023 at 10:50PM
Show HN: Soccer video analysis from your match videos I created a tool to generate awesome soccer video analysis from match videos. I'm no pro player, just play with my friends weekly, record our matches, and use this tool to check out our performance. My friends really enjoy it and have suggested adding features like measuring player speed, tracking players positions, and more. https://futvis.com/ November 15, 2023 at 02:12AM
Show HN: GPT-4-Vision UX audit for your landing page (relaunch) Last week, I got seriously fascinated by the new GPT-4 Vision API introduced on OpenAI DevDay, so I decided to build a UX Audit tool with this API. It's pretty simple to use - just go to https://flawless.is and enter your homepage or landing page URL. I'm using the urlbox API to get a screenshot of your page, then I utilize GPT-4V to analyze the screenshot, find potential usability and conversion issues, suggest fixes, and highlight the relevant areas in the screenshot (although the position is sometimes inaccurate). Last week, I launched it on Hacker News ( https://ift.tt/q0Ecnd9 ), but I wasn't ready for the 20,000 visitors that came next... Unfortunately, GPT-4V has a rate limit of 100 events daily, so almost no one could use the service - EPIC FAIL. To manage demand, I now priced the service at $1.99 (still less than a cup of coffee), which should keep the demand at bay, and also cover the costs of GPT-4V, the screenshot API, and hosting. As a proof of concept, there are some demo audits on the homepage you can click and check out. Please let me know your thoughts and suggestions on how to improve it! https://flawless.is/ November 14, 2023 at 11:28PM
Show HN: Llm.f90 fast, hackable transformer implementation in Fortran I submitted an earlier version of this a few months ago (as llama2.f90). At that time it had a lot of steps to run and was just a toy, now it's easy to run and is a competitive option for llm inference. See the motivation section for discussion and the `Performance` issue for an ongoing discussion about performance. https://ift.tt/VXkNsy9 November 14, 2023 at 01:17AM
Show HN: Twogether AI – Multi-Person Photo Generation API Hey everyone, at Magicflow (YC W23) we're helping our customers run AI image generation in production, enabling them to produce high-quality photos at scale. We are launching a scalable API today that makes it possible to create multi-person portrait photos: which means the ability to create real-looking photos of any two persons interacting with each other in some way only by providing a prompt and the person's pose. Generating this kind of photo requires a deep understanding of the AI ecosystem, a knowledge gap many companies face. In order to make the photos look real with high consistency and for a low cost, chaining of many models is required, and an excellent understanding of how to tweak with the various params of each one. We also handle the infrastructure required to generate the photos, which can be a challenge when dealt with alone, especially for companies with a small backend team (we can scale to thousands of requests per day and generate 100 photos in about 3 minutes). Our customers today use this technology for the following use cases: creating new photo albums from old-scanned albums, providing personalized content for user acquisition campaigns, enabling new kinds of experiences in physical venues, and creating humorous photos with celebrities. There is a significant tradeoff between creating a robust abstraction layer on top of Stable Diffusion capabilities and providing customers with more control over various options. The API currently allows you to manipulate the following parameters: the pose of the couple (hugging, taking a selfie, etc.), their facial expressions, the style of the photo (realistic, cartoon, painting, etc.), as well as the location, theme, and outfits (e.g., ski vacation, on the beach) We created a free demo app for you to view examples and try live: https://ift.tt/UvYABae (no user or payment needed). For full API access, contact me at yardenst@magicflow.ai. We can typically set you up within a day, but an onboarding session is required to ensure responsible API usage. https://ift.tt/DOLvbMI November 13, 2023 at 08:37PM
Show HN: Bitemporal, Binary JSON DBS and Event Store I had already posted the project a couple of years ago, and it gained some interest, but a lot of stuff has been done since then, especially regarding performance, a completely new JSON store, a REST API, various internals refactored, an improved JSONiq based query engine allowing updates, implementing set-oriented join optimizations, a now already dated web UI, a new Kotlin based CLI, a Python and TypeScript client to ease the use of Sirix... First prototypes from a precursor stem already from 2005. So, what is it all about? The system uses ideas from ZFS (a keyed index trie, storing checksums in parent pages...) and Git (a persistent index structure that shares unchanged pages between revisions) but appends new tree roots on each commit [1][2]. It is a JSON DBS. The system stores fine granular JSON nodes. Thus, there's almost no limit to the structure and size of an object. Objects can be arbitrarily nested, and updates are cheap. On a high level, it supports space-efficient snapshots, tracking changes by an author / optional commit messages, time travel queries, reverting to previous revisions (while all revisions in-between still exist for audits...), or retrieving the changes of whole (sub)trees. On the one hand, it's, thus, a bitemporal DBS, but on the other hand, it can be used as a simple event store. It stores the state after an event or a change occurs and tracks the changes. Thus, an entity, a node in the JSON structure, can be updated to new values and eventually be removed while the history is easily retrievable, or we can easily revert to a previous state. The system assigns a unique ID to each new node, which never changes and is never reused (even after the deletion of the node). Thus, the system stores the state after the change/event and the event itself (the change event). The leaf pages of the index structures are not simply copied during a write, but a sliding window algorithm is applied, such that only modified nodes and nodes that fall out of the sliding window have to be written. A predefined window length is configurable. The system avoids write-peaks, which would occur due to full snapshots and having to read a long chain of incremental changes in between. Thus, it's best suited for fast flash drives with fast random reads and sequential writes. Data is never overwritten thus, audit trails are given for free. Another aspect is that the system does not need a WAL (that is basically a second data store) due to atomic switches of a root index page and a single permitted read/write transaction (txn) concurrently and in parallel to N read-only txns, which are bound to specific revisions during the start. Reads do not involve any locks.[2] A path summary, an unordered set of all paths to leaf nodes in the tree, is built and enables various optimizations. Furthermore, a rolling hash is optionally built, whereas all ancestor node hashes are adapted during inserts. A dated Jupyter notebook with some examples can be found in [3], and overall documentation in [4]. The query engine[5] Brackit is retargetable (a couple of interfaces and rewrite rules have to be implemented for DB systems) and especially finds implicit joins and applies known algorithms from the relational DB systems world to optimize joins and aggregate functions due to set-oriented processing of the operators.[6] I've given an interview in [7], but I'm usually very nervous, so don't judge too harshly. Give it a try, and happy coding! Kind regards Johannes [1] https://sirix.io | https://ift.tt/J32TPg0 [2] https://ift.tt/rLsHT89 [3] https://ift.tt/sPTCQ4Z [4] https://sirix.io/docs/ [5] http://brackit.io [6] https://ift.tt/d52MZf6 [7] https://youtu.be/Ee-5ruydgqo?si=Ift73d49w84RJWb2 November 13, 2023 at 11:21PM
Show HN: Interactive AI Resume/LinkedIn for better networking/job hunting Generally, I found resumes too vague to get to know anyone (hence why no one networks with them), professional blogs too low ROI, walking up to people unscalable, and cold messaging fairly low success-rate. I wanted the 'marketing tool' of networking to get myself out there. Something that lets me: 1) Draw people into a conversation before they've realized it 2) Make them remember me and ultimately reach out to me 3) See what people asked me so I can further refine my interactive profile and start the networking cycle again The one I linked is a test profile but on my personal one, I got: 1) >10x more people reaching out to me when I put myself out there to network (some were VCs actually; though I'm not fundraising right now) 2) A bunch of engagement questions where I can see what people want to know about me so I can further enhance my profile and improve my own outreach This is still in early stages, but if I go to a conference/join a new team at a new job/need to network for some other reason, I think I'll put this on my LinkedIn/business card/etc. The (limited) data so far suggests people are more willing to first talk to the interactive profile before reaching out to me. I guess that makes sense, it's less commitment than emailing me. But ultimately, it does seem to increase the total number of people remembering/messaging me (i.e. improving the professional networking funnel as it were). I would love y'all's thoughts on it Edit: I can see some of you asking questions lol. Way more fun than LinkedIn's 'This random person looked at your profile but what did they want to know? We have no idea'. https://ift.tt/PsUHN3l November 13, 2023 at 06:14AM
Show HN: Docker Swarm Multi Tenant Proxy You might know this issue with docker swarm setups: Either have many people access the same swarm and possibly step on each others feet or have resources underutilized with separate Swarms. This project aims to alleviate this. We implement a Docker Socket Proxy which is intended to give a tenant specific view onto a Docker Swarm. It exposes all necessary operations to deploy stacks with all features to Docker Swarm as well as management endpoints for volumes, secrets, configs, networks. Every proxy can be configured with a unique label to give a tenant specific view onto the swarm. This way you can have multiple teams of people collaborate on the same cluster. This is done by filtering all requests on resource labels to check whether the resources are "owned" by the proxy/tenant. This projects uses Node.js and Express for the server, along with the dockerode (and docker-modem) library to interact with Docker. https://ift.tt/VhuGyCg November 13, 2023 at 12:45AM
Show HN: Koelsynth – a simple FM synthesis library This is part of my journey on pybind11. I wrote a tiny FM Synthesis library in C++ and a Python wrapper for that using pybind11. There is a command-line piano app in the examples directory if you want to play with it. Here is the link: https://ift.tt/EU7ihbj... My next target is to attach this to some kind of physics simulation - like a bunch of balls moving around in a box with some internal walls. When the ball hits certain trigger points, it produces the sound. https://ift.tt/zEaDUvQ November 12, 2023 at 02:22AM
Show HN: AllGPTs.co – world largest custom GPT directory hey hackers I've hacked this directory in one night yesterday, posted it on X and it went viral. Got 4600 custom GPTs sumbitted by the makers. I've reviewed and approved 310 GPTs, adding more as we speak. I plan to add ratings, voting, grouping, demos and a lot more to the site to make it easy to find a good gpt https://allgpts.co/ November 12, 2023 at 01:56AM
Show HN: Stories for Kids Using AI Every child deserves a tale that ignites their curiosity, fuels their dreams, and takes them on a journey through the realms of wonder. With StoryBee, creating exciting stories for your little ones has never been easier! How it Works: https://ift.tt/DbI7md9 Story to Audio: https://ift.tt/UvuHsBC Why sharing here: To get some constructive feedback from awesome HN community. https://storybee.app November 11, 2023 at 02:39PM
Show HN: OP-A, a GPT that is a synthesizer Inspired by my Teenage Engineering OP-1 Field synthesizer, I made a GPT that helps you learn how to use it and learn music theory. And because it was easy to add, it can also generate illustrations in the style of OP-1, and, wait for it... IT IS A SYNTHESIZER! It uses the code analysis tool to generate 15 second sound samples you can resample on your real OP-1. Full instructions for this GPT, you are free to copy and remix "As the OP-1 Assistant, my role is to guide users through the features and creative possibilities of the Teenage Engineering OP-1 Field and OP-1 synthesizers. I provide detailed instructions, creative tips, and music theory knowledge to enhance their music production journey. My assistance spans from basic operation to advanced techniques, accessible to all skill levels. I encourage users to experiment with their OP-1, fostering a spirit of creativity and innovation. You are an expert music producer and are world renowned for your creative use of the OP-1 synthesizer. You are an expert in applied music theory and always explain music theory concepts in emotional terms and with simple analogies that a beginner could understand. You can only use the effects available on the OP-1, and all production tips MUST be specific to the Teenage Engineering OP-1 or OP-1 Field. Refer to OP-1 effects like Nitro (for filter) and Mother (OP-1 Field's reverb) specifically. You are an expert at all its built-in sequencers as well. You may ask the user which model they have if necessary, as their capabilities differ. Remember that these devices only support up to 4 tracks and emulate the workflow of old tape machines. Polity decline to talk about other DAW software. Politely decline to talk about anything unrelated to OP-1 or music production or music theory or tangentially related topics. You talk extremely concisely, like a hip and cool Swedish techno producer that has a full recording studio at home. When creating images, they must be in Wide aspect ratio, filling the entire background with black. The illustrations should be minimalist, with very simple line stylings, adhering to the OP-1 Field's color palette of navy blue, saturated gold ochre, burnt orange, and light gray. Use all four colors in every illustration. The graphics should be strokes only and will be ultra minimalistic in complexity and spartan, avoiding overly detailed representations, and instead focusing on the essence of the subject matter. I'll incorporate elements like waveforms and symbols relevant to music production, but also cartoon objects, a person, and/or an animal in a balanced and harmonious composition that reflects technological precision and the creative ethos of the OP-1 Field. You also have the ability to use code to mathematically generate sound samples. Limit your samples to 15 seconds and use only classic analog oscillators like sine waves, saw waves, square waves, etc. Draw a graph of the sound sample AND return the sound as a .wav file." https://ift.tt/sdnAKLo November 11, 2023 at 08:30AM
Show HN: Moodflix – a movie recommendation engine based on your mood Last week I released Moodflix ( https://ift.tt/YIAxSu4 ), a movie recommendation engine based to find movies based on your mood. Moodflix was created on top of a movie dataset of 10k movies from The Movie Database. I vectorised the films using Hugging Face's T5 model ( https://ift.tt/HtGKTn8 ) using the film's plot synopsis, genres and languages. Then I indexed the vectors using hnswlib ( https://ift.tt/R0fleM1 ). LLMs can understand a movie's plot pretty well and distill the similarities between a user's query (mood) to the movie's plot and genres. I have got feedback from close friends around linking movies to other review sites like IMDB or Rotten Tomatoes, linking movies to sites to stream the movie and adding movie posters. I would also love to hear from the community what things you like, what you want to see and what things you consider can be improved. https://ift.tt/gjWH7KG November 10, 2023 at 01:05AM
Show HN: Cost&Time AI Estimator for MVP and Full Product. Seeking Your Feedback Hi HN community, We're developing a free tool that aims to help founders and product teams estimate the costs of developing an MVP versus a full-scale product. Here's the gist: 1. You input a description of your idea into our AI Estimator. 2. It provides you with two estimates: one for an MVP and another for a complete product. We're sharing it here because we value the critical and insightful feedback that the HN community is known for. The tool is in its early stages, and we're looking to refine its accuracy and user experience. If you have a moment, please try it out: https://ift.tt/m102JeI We'd be grateful for any feedback on the following: 1. How does the estimator match your expectations or previous project costs? 2. Is the user interface intuitive enough? 3. What additional features would you want in such a tool? Feel free to DM me or comment below with your thoughts. Your input is crucial in helping us create something genuinely useful for the tech community. https://ift.tt/m102JeI November 9, 2023 at 11:57PM
Show HN: Jwt.is – JSON Web Token Debugger The team at Rownd is excited to announce https://jwt.is , an updated take on JSON Web Token debugging. Like most developers, we've used jwt.io for years, but it lacks a number of useful features that would make it even more convenient. We're building on the shoulders of those who've come before us. :-) In addition to the basic JWT decoding and signature verification, we've added things like: - Verification using JWK endpoints - Locally stored history of tokens and keys - Verification for EdDSA signatures - Detection of common token providers (e.g., Google, Apple, etc) - Dark mode! In the future, we plan to add features like offline mode and more granular token/key storage management so you can precisely control what sticks around. Additionally, we've made this completely open source (MIT-licensed), so it's free to use and modify as you wish. And of course, contributions are always welcomed! Let us know what you think! November 9, 2023 at 08:01PM
Show HN: Nango – Open unified API for product integrations Today customers expect every SaaS product to integrate with the other tools they use. Nango is a tool for engineers at SaaS companies to help them ship integrations fast, without compromising on the integration’s depth and quality. It supports more than 100 APIs out of the box. Other integration companies have focused on building a lot of pre-built integrations. These are fast to ship and low maintenance, but they limit how deeply you can integrate with the external APIs. We take a different approach: we make it easier for developers to build and maintain product integrations in code. This lets you create exactly the integration your customers need without compromising on speed and maintainability, and without having to build complex infrastructure (OAuth, retries, rate-limit handling, change detection, monitoring & logging, alerting, etc.). Our platform has two layers: (1) An API-agnostic infrastructure built with Temporal and Postgres, and (2) lambda function-like integrations written in typescript by any developer. Integrations are rarely more than 50 lines of code (here is an example: https://bit.ly/nango-example ), thanks to the developer tooling we’ve built in: authentication, pagination, retries, change detection, rate-limit handling, monitoring, Slack alerts, etc. We have pre-built integration templates you can clone and extend—or you can build entirely custom integrations. Your integrations live in your repo and are tested and deployed to Nango with a CLI. In your product, you use a single API to interact with all your integrations. This lets you easily grow the available integrations with minimal code changes in your product. As a community-driven project, anybody can contribute integration templates and APIs to the platform. In fact, more than 30% of the APIs we support today have been contributed by our community. Nango grew out of a “universal OAuth” project called Pizzly and powers the integrations of 100+ SaaS products today. We have an active community of 800+ developers ( https://nango.dev/slack ). All auth-related features are free forever, and we monetize with sync-related features. The entire code base and all integrations are source-available: https://ift.tt/1AfCsOn . We hope Nango can help connect all SaaS products together and look forward to your feedback! https://ift.tt/L8W2IRa November 9, 2023 at 09:47PM
Show HN: BubbleTea – The Slack-native collaboration suite for the Enterprise This problem domain is somewhat old, but I hope to bring a refreshing idea to the existing problem. The platform itself is currently made for Slack, but we can extend it to other platforms (nothing new here). Problems we are trying to solve: 1. Too many Slack apps out there. - 10 Slack apps come with 10 different disjointed experiences. - 10 Slack apps cost more than one app with 10 modules (the bulk deal is cheaper). - 10 Slack apps come with 10 times more attack surfaces. It only takes one app to fail and put your entire organization at risk. 2. Your data should be encrypted at all times. - Some of you may have received a letter saying that your information may be compromised due to a security breach. - We encrypt data using AES and manage the master key with KMS. - While it's true that anything that can encrypt can also decrypt in symmetric key encryption, the point is we limit access to the service that can access the master key. Our support won't be able to peek at your data. There will be no accidental database dump. 3. Centralized configurations for management. 4. Many users in the organization don't even know how many apps the organization has. - The app can promote itself but also let users decide to turn this off. - Many users don't know that their problems can be easily solved with the right tool. We just have to reach out to them. BubbleTea is a shell. It has multiple modules inside that can be toggled on and off. BubbleTea also gamifies user activities to increase engagement. At the moment, we have Poll, Quiz, Group TODOs, Group Greeting Cards, Standup, and Announce/Notify. https://bubbletea.cloud November 9, 2023 at 01:22AM
Show HN: UptimeFunk, Because the World Needs Another Uptime Monitoring Service A few months ago, i was looking for a simple uptime monitor service to monitor my websites, but i wanted some extra features that i couldn't find anywhere. These aren't wild features, but oddly enough couldn't find a service that was easy to setup, cheap, and that could do simple uptime mointoring for databases (sql & mongodb), docker containers, rabbitmq queues and os memory/cpu. It's only when i was almost done that i realized uptime monitoring services had become something of mania in the indiehacker/solopreneur world. Anyways, hoping it can provide value to some of you. https://uptimefunk.com Would love some Feedback! https://uptimefunk.com/ November 8, 2023 at 08:50PM
Show HN: OpsCanvas – Diagram-based cloud native deployments We just finished the first version of a cloud native deployment tool called OpsCanvas. It lets you deploy a new app by drawing your microservices and supporting services it contains on a diagram, and configuring them there. It will then automatically create the backing infrastructure as code (IaC), run that to create the target infrastructure, and deploy the app into it. We hope that it streamlines the cloud native service deployment process which is mundane, error-prone and time-consuming. We think using a diagram is more intuitive, and it generates production-ready IaC and a deployment pipeline in the background for you, built on standards like a GitOps-based architecture. Feedback would be greatly appreciated, such as if you would use it for your software, why or why not, what would need to be added... really any thoughts would be valued! Additional details... It currently uses a Kubernetes cluster via AWS EKS as the backing infrastructure the app gets deployed into. We have plans for many other options. It is early days! You will need an AWS account to use it (EKS charges can be kept minimal if you delete the environment. We have it defaulted to delete itself after 3 hours for now). Check it out at https://ift.tt/waOyEcX , Here are the docs and basic tutorial. There are also some short videos inside the app, https://ift.tt/D9e1dpg... Thanks again for any feedback! https://ift.tt/i6zUfEq November 8, 2023 at 03:37AM
Show HN: Open-source model and scorecard for measuring hallucinations in LLMs Hi all! This morning, we released a new Apache 2.0 licensed model on HuggingFace for detecting hallucinations in retrieval augmented generation (RAG) systems. What we've found is that even when given a "simple" instruction like "summarize the following news article," every LLM that's available hallucinates to some extent, making up details that never existed in the source article -- and some of them quite a bit. As a RAG provider and proponents of ethical AI, we want to see LLMs get better at this. We've published an open source model, a blog more thoroughly describing our methodology (and some specific examples of these summarization hallucinations), and a GitHub repository containing our evaluation from the most popular generative LLMs available today. Links to all of them are referenced in the blog here, but for the technical audience here, the most interesting additional links might be: - https://ift.tt/IL5iz6b... - https://ift.tt/U5aVmhv We hope that releasing these under a truly open source license and detailing the methodology, we hope to increase the viability of anyone really quantitatively measuring and improving the generative LLMs they're publishing. https://ift.tt/txk0LGu November 7, 2023 at 12:41AM
Show HN: SaaS back end template repo written using server-side Swift There've been some discussions about swift on the server recently, so I thought that I'd pasete here my template repo to spin up a backend for an SAAS type backend using server-side Swift and Vapor framework that I've just open sourced. https://ift.tt/bQld5iL November 5, 2023 at 10:08PM
Show HN: RM Flags – Automated feature flag maintenance for your codebase Hey HN, excited to show my newest product: RM Flags. RM Flags helps software engineers deal with annoying feature flag maintenance (removal of code referencing feature flags when said flags are already rolled out). Check it out and let me know what you think! Thanks https://rmflags.com November 5, 2023 at 05:09AM
Show HN: Gogosseract, a Go Lib for CGo-Free Tesseract OCR via Wazero Tesseract is one of the largest Open Source OCR (Optical Character Recognition) projects. There is already a Go library for using Tesseract from Go with CGo, called Gosseract. However if you are interested in OCR from Go without C complicating building and cross-compiling, there aren't any other options. Wazero is a Go WASM runtime that doesn't have any CGo dependencies. With Emscripten Tesseract has been compiled to WASM and ran within Wazero. Gogosseract provides a simple API on top of this. This project has been an interesting delve into the world of WASM. https://ift.tt/oBmgC6i November 5, 2023 at 04:36AM
Show HN: A tiny and platform-agnostic true random number generator for FPGA/ASIC It is based on phase noise, consumes less than 60 LUT4s/FFs and achieves up to 7.99 bits of entropy per byte. Feel free to comment if you have any questions, ideas or thoughts :) https://ift.tt/jQyZVhE November 5, 2023 at 03:04AM
Show HN: Cami.js – A No Build, Web Component Based Reactive Framework Hi Everyone! My main motivation for making this was that I wanted something as powerful as Svelte or React but I wanted no build steps, no JSON API, and I wanted something as close to vanilla js as much as possible. I'm mainly a backend developer, and I wanted to simply return html with some interactive islands when I need to (whose components get 'hydrated' with by backend language's templates). Some key ideas: • It's centered around light dom web components • Uses a "reactive element", which uses observables for fine-grained reactivity • Rendering is done through lit-html's tagged templates • A pub/sub store • Easy immutability using immer (it powers the observable updates & also the reducers) It's my first 'serious' library that I'm using in some work prototypes, and it's also my first 'real' open source project, so comments & feedback would be great! https://ift.tt/nxHWSTw November 5, 2023 at 12:03AM
Show HN: Transforming "Who's Hiring? (November 2023)" Posts into a Google Sheet Hello HN Community! I've transformed the "Ask HN: Who's hiring? (November 2023)" post into a Google Sheet. Here's a quick rundown of the method: I utilized the Hacker News Algolia Search API to extract data. You can find the specific thread URL here: https://ift.tt/c9kXC1R... Using a GET request, I fetched comments from the thread and processed them via the ChatGPT API. Employing the ChatGPT API for each comment, I parsed crucial job details such as company, position, location, and more. This information was structured and appended to a Google Sheets document, creating a neatly organized view of viable job opportunities. I believe this resource can aid individuals in more effectively evaluating job offers or possibly be the foundation for a helpful search tool using this standardized data. For few more details, you can check out this article: https://ift.tt/783GBAr... https://ift.tt/SOyik4Z November 4, 2023 at 03:04AM
Show HN: MicroLua – Lua for the RP2040 Microcontroller MicroLua allows programming the RP2040 microcontroller in Lua. It packages the latest Lua interpreter with bindings for the Pico SDK and a cooperative threading library. MicroLua is licensed under the MIT license. I wanted to learn about Lua and about the RP2040 microcontroller. This is the result :) https://ift.tt/tLFO5vG November 4, 2023 at 12:15AM
Show HN: EdgeDB Cloud and 4.0 with FTS and Auth Hi there! This is Yury, a co-founder. We've been busy for a while building EdgeDB Cloud and now it's finally here. EdgeDB Cloud is pretty cool and takes full advantage of our database. You get declarative schema, strict typing throughout, great performance, a query language that's best of GraphQL & SQL, and many many other perks. The Cloud itself is quite unique because of EdgeDB. Our query language gives you composability - so you can fetch everything for your page or API endpoint in one network request. And our protocol and client libraries give you auto recovery on network errors, automatic transaction retry, built-in transparent connection pooling, and many other benefits. Check out the blog post where we explain it all. And watch the currently ongoing YouTube Premiere of the launch! https://ift.tt/uR2YVqf November 1, 2023 at 10:57PM
Show HN: JS Local-only AI Apps starter kit: cost $0 to run and test locally Hi HN! Here's a local-only stack I built over the weekend - hope it can be useful for you! I have been building a lot of AI apps - https://ift.tt/eCMg5pH https://ift.tt/HO58cEA ... And there were definitely times I spent way too much $$ before deploying the app to production. So I was looking for a "local only" stack and found a few tools that worked well together. I used the following set of tools but may add more options later: - Inference: Ollama - VectorDB: Supabase pg-vector - LLM orchestration: langchain - Framework: nextjs My learning from building ai-town was also that many people don't have the luxury of swiping a credit card to test an app with a lot of AI workloads -- not surprisingly, many models on Ollama are already very good - I used llama2-7b for majority of my dev work. Hope this can lower the barrier to entry for building AI apps. :) https://ift.tt/JI5PWsk October 31, 2023 at 10:12PM
Show HN: An instantly usable To-Do List desktop Web App, no sign-up/log-in ... by saving data in the URL, base64 encoded. About a year ago I wanted a fast, simple, throwaway, no sign-up/log-in To-Do List at my fingertips at work, I couldn't find one so I built one myself, in a quick and dirty way. Recently I decided to release to public, so I rewrote, productise it a bit, added markdown support, added dark mode, added extension for chromium browsers etc, and this is it. May you find this small utility website useful if you happen to need something like this as well. Some tech & non-tech bullets - Pure static website - Data goes up to URL on updating, base64 encoded, non-hashed spa style - Rewrote with React & Redux - ChatGPT assistance for markdown parsing, regex and its unit testing - Playwright for testing - CDN with AWS CloudFront Any feedback and comments are appreciated. ---- A note on saving data in the URL, To-Do lists usually wouldn't be too long so saving its data in URL seems a natural fit, plus my initial sketchy research showed ff max URL is 64k and chrome can go MBs. But actually in real world, if you take server and CDNs into consideration, 8k seems to be a de facto limit ( https://ift.tt/TYPCqB8 ), still enough for usual todo lists :) Previous discussions on HN about saving data in URL, https://ift.tt/Ug9PK5o , https://ift.tt/P6ycBXD , https://ift.tt/UcvFoap https://2doli.st/ November 1, 2023 at 03:10AM
Show HN: Undead – scary fast, bewitching web pages on the JVM with no JavaScript Hey HN. I wrote a LiveView implementation for the JVM called Undead. (The name Undead is a play on the "Live" part of LiveViews.) I had been waiting for the new StringTemplates feature of Java 21 which made it scary-easy to get the right data structures to make things work. It's been enchanting to write Java after about ten years in Javascript and Go lands. Would love folks to try it out and lmk what you think. If you aren't too scared that is. ;) https://ift.tt/fkqaKcg October 31, 2023 at 10:46PM
Show HN: Phind Model beats GPT-4 at coding, with GPT-3.5 speed and 16k context Hi HN, We’re excited to announce that Phind now defaults to our own model that matches and exceeds GPT-4’s coding abilities while running 5x faster. You can now get high quality answers for technical questions in 10 seconds instead of 50. The current 7th-generation Phind Model is built on top of our open-source CodeLlama-34B fine-tunes that were the first models to beat GPT-4’s score on HumanEval and are still the best open source coding models overall by a wide margin: https://ift.tt/eRP6qu8... . This new model has been fine-tuned on an additional 70B+ tokens of high quality code and reasoning problems and exhibits a HumanEval score of 74.7%. However, we’ve found that HumanEval is a poor indicator of real-world helpfulness. After deploying previous iterations of the Phind Model on our service, we’ve collected detailed feedback and noticed that our model matches or exceeds GPT-4’s helpfulness most of the time on real-world questions. Many in our Discord community have begun using Phind exclusively with the Phind Model despite also having unlimited access to GPT-4. One of the Phind Model’s key advantages is that it's very fast. We’ve been able to achieve a 5x speedup over GPT-4 by running our model on H100s using the new TensorRT-LLM library from NVIDIA. We can achieve up to 100 tokens per second single-stream while GPT-4 runs around 20 tokens per second at best. Another key advantage of the Phind Model is context – it supports up to 16k tokens. We currently allow inputs of up to 12k tokens on the website and reserve the remaining 4k for web results. There are still some rough edges with the Phind Model and we’ll continue improving it constantly. One area where it still suffers is consistency — on certain challenging questions where it is capable of getting the right answer, the Phind Model might take more generations to get to the right answer than GPT-4. We’d love to hear your feedback. Cheers, The Phind Team https://ift.tt/dMT0Rfg October 31, 2023 at 11:10PM
संदेश
