Show HN: Fill and Sign PDF from any website with a single line of code https://ift.tt/Zl0kHz9 December 30, 2023 at 09:33PM
Show HN: Intuitive nutrition information Hey everyone, I've been building this nutrition tracker and calorie counter recently, after being frustrated by existing products for ages. I built a similar app 8 years ago [1], but came back to this problem again since there are still no good solutions here. Lmk your thoughts and improvement ideas :) [1] https://ift.tt/R3qkBEn https://spe.lt/ December 30, 2023 at 01:48AM
Show HN: Alarmclocktab.com – phone's dead and I need an alarm, stopwatch, timer Web based time tools including alarms, timers, stopwatch, and time zone converter. Resources also include Tabata, Pomodoro and other useful timers, tips and tactics for managing time and time tracking. Dark mode included! https://ift.tt/Sw0qgc5 December 29, 2023 at 11:03PM
Show HN: Calcium Imaging Pipeline Tool OptiNiSt(Optical Neuroimage Studio) is a GUI based workflow pipeline tools for processing two-photon calcium imaging data. OptiNiSt helps researchers try multiple data analysis methods, visualize the results, and construct the data analysis pipelines easily and quickly on GUI. OptiNiSt's data-saving format follows NWB standards. OptiNiSt also supports reproducibility of scientific research, standardization of analysis protocols, and developments of novel analysis tools as plug-in. https://ift.tt/kHKQS1c December 29, 2023 at 03:04AM
Show HN: workin – A web app to create and run smart trainer workouts I used TrainerRoad for my indoor bike workouts during Winter. Its subscription costs $20 per month, which seemed unreasonable for the one thing I was using: the workouts. So I decided to create an free and open-source alternative. I've been using it for almost a month now and it feels great to have built something useful for once :) https://ift.tt/J9hvOYl December 29, 2023 at 12:26AM
Show HN: Scan QR codes to check-in guests registered via Google Forms™ Hi HN! I made a no-code platform for creating physical data collection apps, using QR codes [1]. It does not yet have a self-service config UI though, which limits adoption. That's why I recently released a Google Forms™ add-on for QR code check-in, based on the platform. This focused use-case makes it easy to provide a fully self-service config UI. How it works: 1. Create your Google Form as you normally would [2] 2. Activate the add-on if you hadn't already [3] 3. Craft a confirmation email to be sent to each form responder Upon each form submission, the add-on will send a PDF with a unique QR code (a V4 UUID) to the responder. Have guests present this code at the event, and record check-ins in bulk using the included QR scanner. See here [4] for more information, or try the Google Sheets™ version [5] (which doesn't send email). [1] https://ift.tt/viA2WKE [2] https://ift.tt/YhmAqPV [3] https://ift.tt/bTmHzNy... [4] https://ift.tt/iedjOwv... [5] https://ift.tt/suKQzJn... https://ift.tt/C4LjtuZ December 28, 2023 at 12:43AM
Show HN: Meal planning app I built after losing 50kg myself in 2022 1500cals is a calorie and excercise tracking tool that I built after losing 50kg myself in 2022 through simply watching what I ate and starting to exercise. The main "trick" I uncovered is simply to plan meals ahead. It sounds boring, it sounds annoying, but really its such a tiny effort for such outsized and life changing rewards that it should be a no-brainer. The existing tools I found were more focused on tracking after eating which is great in terms of capturing data and measuring over time, but I found really didn’t help me with locking in a routine. I wanted to fix this and bring some of this planning side along with tracking into an easy to use app. This is 1500cals! https://1500cals.com December 27, 2023 at 10:14PM
Show HN: Try or Join Aisbreaker.org: Open API to Many AIs (ChatGPT, HF AIs,) AIsBreaker offers a developer-friendly API for streamlined access to a variety of generative AI services, including OpenAI/ChatGPT, all Hugging Face AIs, and more ... to provide simplicity and to avoid vendor lock-in. Demo: JavaScript code: https://ift.tt/k7fwyvW Demo: Live web app: https://ift.tt/Szumq5x Collaborators wanted: https://ift.tt/wcP7gDO https://aisbreaker.org/ December 27, 2023 at 03:52AM
Show HN: Made a batching LLM API for a project. Mistral 200 tk/s on RTX 3090 I was running into an issue with a vLLM bug that affected multiple GPUs and I needed a stand-in while that bug was getting fixed that used the same API format but had better performance than the API on text-generation-webui. It's very rough. I'm not a coder by trade. But it's very fast once you have many simultaneous connections. https://ift.tt/OUFapfo December 27, 2023 at 01:22AM
Show HN: I made an API to manage the cloud storage I have freelance projects involves the creation and administration of Google Cloud buckets. However, managing access permissions can be a bit challenging at times. To streamline this process, I've decided to develop an API to manage all my buckets, each associated with its unique API key. I've created a demo to showcase the capabilities of this solution https://scroud.app December 25, 2023 at 11:41PM
Show HN: I Made a Christmas Quiz For the last couple of years I've created a Christmas quiz, which you can read about and play here: https://ift.tt/WiFMefD... . This year I created another quiz, this time general Christmas questions. Like last year the tech I used: - Astro - React - TypeScript - Framer Motion for the animation - A Netlify function that saves and loads the scoreboard from Azure Table Storage Unlike last year which was somewhat easy to cheat on by Googling, this time, the longer you take to answer a question the fewer points you get. I hope you have fun and a great Christmas https://ift.tt/Xm3QANU December 25, 2023 at 05:31AM
Show HN: Boarda – Focused Offline Kanban Board I like using Kanban boards, so I created a program to manage them using Zig and raylib. You can check it out here https://ift.tt/rUck2w9 It can open multiple boards, has undo/redo, clipboard support, and many other features. You can read about in the page above or try it yourself. https://ift.tt/rUck2w9 December 25, 2023 at 09:07PM
Show HN: Capit – Social Media Image Caption Generator Hey HN! I just recently released my latest iOS App, Capit - it is a simple app built to generate captions for social media posts, all directly from images & video, utilizing gpt-4-vision. Users just choose an image, select a tone, and get 5 great caption options. My latest update adds Profiles to allow you to save your settings for different social media profiles, along with a simple caption editor to let you customize your captions further. The app is a subscription service - but lets you try it for free if you just hit the X button on the paywall. I built the app as my wife runs a small business for which Instagram is a huge traffic generator...but she often gets stumped at what to write in terms of captions, especially when we've been open for years and it feels like you are just recycling content in many cases. Capit isn't really meant to automatically post for you - it is a tool for improving your own creativity. I'd appreciate any & all feedback! https://capit.app December 24, 2023 at 12:49AM
Show HN: Extension that combines multiple tools into one I built Browser Boost as an alternative to having to trust multiple extensions from various developers, which are often closed source, collect user data, and come with undocumented permissions. Browser Boost collects no user data, does not require signup, has no servers, is entirely open source, and is 100% free. Tools: - Boost Volume: Increase the maximum volume of your browser up to 600%. - Change User Agent: Spoof user agent to emulate different devices or browsers. - Enable Right Click & Select: Force enable right clicks, copy, & text selection on sites that have disabled right click or selection interactions. - Save Image as Type: Add context menu to save images as JPG, PNG or WebP. - Disable WebRTC: Disable WebRTC requests to prevent IP address leaks. Chrome Download: https://ift.tt/tOcNS8U... (Currently working on the Firefox version) Github: https://ift.tt/8S3OyIl https://ift.tt/8S3OyIl December 23, 2023 at 10:23PM
Show HN: A chess platform built for teams only I've been working on this pet project for almost a year now. Finally at a point where I'm ready to share it a bit more openly. I've played with family and friends mostly and so far have received positive feedback. Want to get some honest real world feedback now. Let me know what you think. If you like chess this is for you! https://ift.tt/jyt9xKn December 22, 2023 at 10:55PM
Show HN: I'm open-sourcing my game engine Modd.io is a collaborative game editor that runs in browser. It's kind of like Figma for game dev. We made this engine low-code and multiplayer-first, so developeres can quickly prototype casual multiplayer games. I hope some of you guys will find this useful. Would love to hear feedback also. Thank you. Engine Demo: https://www.modd.io https://ift.tt/FjPMfGl December 20, 2023 at 01:08PM
Show HN: Live-ish illustration for TTRPG campaigns My D&D group typically casts music to our TV during sessions, and very occasionally, the default background image lines up nicely with what the party is doing (e.g. a picture of a mountain while we’re exploring a mountain pass). Using Whisper speech-to-text, GPT 3.5, and DALL-E 3, I rigged up something to make that happen automatically. The results are certainly interesting. The images generated are typically much closer to what’s actually going on than the previous example, but get many of the small details wrong. https://ift.tt/WKmtTfO December 21, 2023 at 10:14PM
Show HN: JungleReader – A free and open-source web reader Hello! I built a web-app called JungleReader that lets you follow RSS, Mastodon, YouTube, Reddit, and random blogs, all in one place. I made JungleReader as a central place to follow the blogs and sites I like, which I otherwise often lose track of. My next project (which is WIP) is a web-app called JungleWriter where you can build a simple blog and export it for hosting. I love finding a good, free-standing blog, and hope to make it easier for people to read and write them. Feedback is appreciated, thanks :) https://ift.tt/b3xMzSm December 21, 2023 at 03:45AM
Show HN: Slack knowledge curator to extract FAQ from discussion threads Hi HN! Excited to share Snowshoe, a Slack bot that transforms your Slack threads into a searchable StackOverflow-style Q&A. Snowshoe automatically curates key discussions from Slack, creating a dynamic, easy-to-navigate knowledge hub. A lot of technical discussions among engineers happen in Slack threads. Yet the unstructured thread makes its inherent knowledge difficult for search and thus the discussion insights aren’t often well documented & maintained. LLM is surprisingly good at summarization & extraction. Thus I think it’s good timing to give it a try. Currently Snowshoe focuses on knowledge extraction in Q&A format. What’s next? I feel like there would be a lot of value in the auto-curated Q&A, e.g. auto question answering, knowledge hub, Slack archive for SEO. Happy to hear your thoughts! https://snowshoe.dev/ December 21, 2023 at 12:21AM
Show HN: Command line tool for extracting secrets from WARC (Web ARChive) files Troll-A is a command line tool for extracting secrets such as passwords, API keys, and tokens from WARC (Web ARChive) files, such as the ones from the Common Crawl project or the Archive Team. https://ift.tt/mloU8Kp December 20, 2023 at 10:50PM
Show HN: Quack Companion – VSCode extension for OSS contribution assistance Hello there, I'm FG (short for François-Guillaume)! I’m building Quack Companion ( https://ift.tt/w3tBx06 ), a tool designed to bridge the gap between project maintainers and contributors in open-source software (OSS). If you've ever contributed to OSS, you're familiar with the challenge: diving into a new codebase, aligning with maintainers' expectations, and making meaningful contributions can be daunting. For maintainers, managing a flood of contributions and providing guidance while maintaining quality is overwhelming. As a contributor to PyTorch & an OSS author, I’ve been on both sides over the years. This challenged me to become a better engineer & team player, seeking ways to facilitate smooth collaborative software development. This is where Quack steps in, as your AI companion for software team alignment. For developers, it's like having a seasoned mentor guiding you through the intricacies of a new codebase, offering live in-line hints based on the project’s guidelines to craft high-quality pull requests. For maintainers, it’s a practical toolset to identify and address workflow inefficiencies and align contributions effectively with project objectives. As code generation gets commoditized, aligning these diverse efforts becomes critical if you wish to convert individual productivity boosts into team velocity. Quack AI is here to solve this alignment, and make collaborative software development scale without losing any efficiency. We're committed to keeping the service accessible and free for OSS communities while we plan to generate revenue from the enterprise suite. The platform (React, Next JS), the IDE extension (VSCode) and the backend API (Python, FastAPI) are licensed under Apache 2.0. You can find a short demo here [1] and the GitHub project here [2]. Our roadmap includes: * offering autocompletion and code chat in the IDE, making the contribution process even more intuitive and seamless; * finalizing the transition of the community version to hostable OSS models; * identifying ambiguities or unspecified aspects of a given project’s guidelines; * developing a Fitbit-like feature for your software development productivity to identify bottlenecks in your workflow. This is still the early days, but we've seen how sharing a "public design doc" with the community can significantly improve the outcome! How was your experience of managing inbound as maintainers? What were your personal hacks to mitigate those challenges? We'd love to hear about how it has impacted your developer life or if you have any feedback about the above. Cheers! [1] Demo video: https://ift.tt/uwvKHcB [2] Open Source repo: https://ift.tt/w3tBx06 Our documentation: https://ift.tt/HQm9Tux https://ift.tt/w3tBx06 December 20, 2023 at 10:13PM
Show HN: Userscript to improve the readability of deep nested comments on HN Hi everyone, I love browsing Hacker News comments on my smartphone, especially the very popular posts with 100s of replies. However the deep nested comments can get hard to read if the text is squashed together on the right side of the screen. My userscript enhances the readability by providing a button to expand the comment to full width. https://ift.tt/RlPkYrf December 20, 2023 at 02:01AM
Show HN: Canary Checker – An OSS Kubernetes Native Health Check Platform I am Moshe the Founder @ Flanksource, and today we are releasing canary-checker, an open source, kubernetes native health check platform that provides a unified view of health across the entire stack. Canary checker was born out of the need for platform teams (particularly those running on- premise or on less than perfect infrastructure) to understand the health of applications running on top of the platform. Unlike other solutions such as prometheus blackbox exporter, kuberhealthy, and synthetic testing offered by major cloud providers, our approach goes beyond health collection. Canary Checker enables health aggregation by ingesting alerts from Prometheus, Cloudwatch, Dynatrace, and for running full test suites using tools like K6, Playright, Robot, etc.. This approach provides a unified view of system health without the need to browse through many dashboards. Canary Checker can also replace the need for multiple Prometheus exporters. It empowers you to export metrics from responses across 35+ supported protocols, including HTTP, SQL and Elasticsearch. Something fairly unique to canary-checker is the ability to build stateful checks that use the results of a previous check in subsequent checks to generate metrics around historical data such as logs, logins, and activity stored in db tables. https://ift.tt/P3zRU7W December 19, 2023 at 11:44PM
Show HN: Paradict – Streamable multi-format serialization with schema Hi HN ! I'm Alex, a tech enthusiast. I'm excited to show you Paradict ( https://ift.tt/gruRN3H ), my solution for streamable multi-format serialization. Although JSON, YAML, and TOML are all human-readable, they serve different purposes. For example, TOML is specifically designed for configuration files while JSON is used as a data interchange format. Sometimes an initiative to create a binary version of JSON arises and as far as I know, it ends with an unidirectional mapping of datatypes. There is no silver bullet, yet one coherent solution built from scratch that addresses multi-format (binary and textual) serialization and configuration files would be a step forward. Earlier this year, I accidentally designed a textual data format to represent complex data structures inside a document divided into sections. The project, namely Jesth (Just Extract Sections Then Hack'em), generated an interesting discussion on HN ( https://ift.tt/tSiP5CD ). Out of curiosity, I ran some benchmarks using Jesth, JSON and MessagePack, with and without Gzip compression against a large JSON file downloaded from the web. The benchmarking gave me insights that led to the decision to evolve Jesth's ideas into a new multi-format serialization solution. I designed and built Paradict from scratch to serialize and deserialize a dictionary data structure. Although Paradict's root data structure is a dictionary, lists, sets, and dictionaries can be nested within it at arbitrary depth. A Paradict dictionary can be populated with strings, binary data, integers, floats, complex numbers, booleans, dates, times, datetimes, comments, extension objects, and grids (matrices). There is also a schema-based validation mechanism that can contain programmatic checkers. The binary serialization format is designed with compactness in mind such as Pi with its first two decimal places, the Golden ratio with its first two decimal places, and the date of the funeral of Pope Benedict XVI would each be encoded on two bytes (not counting their respective 1-byte tag which starts each Paradict binary datum). This binary format has two levels of granularity for continuous data stream processing: a datum at the low level, which is in some cases a 2-tuple composed of a tag and its payload, and the message at the high level which is a dictionary data structure. The textual serialization format has two modes: data and config modes. Config mode implicitly treats dictionary keys as strings, removing the need to surround them with quotes, and unlike the colon (:) between a key-value pair in data mode, it uses the equal sign (=) as separator. This textual format has two levels of granularity for continuous data stream processing: a single line of text at the low level and the message at the high level which is a dictionary data structure. Here is a valid Paradict configuration document that contains a "user" section: [user] # no comment id = 42 name = 'alex' birthday = 2042-12-25T16:20:59Z photo = (bin) 54 68 69 73 20 69 73 20 6E 6F 74 20 61 20 70 68 6F 74 6F 67 72 61 70 68 weight_matrix = (grid) 1 0 1 0 0 1 0 1 1 0 1 0 books = (dict) romance = (list) 'Happy Place' 'Romantic Comedy' sci_fi = (list) 'Dune' 'Neuromancer' epitaph = (text) According to the law of conservation of energy, no a bit of you is gone; you are just less orderly. --- Under the hood, Paradict uses Braq ( https://ift.tt/T4geuoB ), the most obvious way to section a document (as shown just above), and Ustrid ( https://ift.tt/2LwcNb9 ), to uniquely generate string identifiers. Paradict is available on PyPI and you can learn more by reading its README, browsing the source code or playing with its tests. Let me know what you think about all this ! https://ift.tt/gruRN3H December 18, 2023 at 10:00PM
Show HN: DALL-E 3 and GPT-4 Vision Recursive Image Generator It's like Gartic Phone for images, powered by AI. I created an automated DALL·E 3 and GPT-4 Vision Image Recursor. Simply provide a starting prompt and see how AI iterates on it. It's perfect for the 'Make it more' trend. You'll need to provide your own OpenAI API Key, but the website is open source! I've linked to the GitHub on the website. Try it out: https://ift.tt/Z9KcIOT https://ift.tt/xTcOYW7 December 18, 2023 at 09:09PM
Show HN: Get any piece of Google Earth as a single normalized glTF 3D model Google released an API in May to get fetch 3D Tiles of anywhere on Earth. Using this in standard 3D engines like Blender is tricky because (1) the tiles are in a geographic coordinate system (2) you get a lot of little tiles at varying quality levels I wanted to simplify this so all you need to do is get an API key, select a map region and a zoom level, and get one combined glTF file that you can throw into any engine. Especially if you're just prototyping and want to see how this data looks in your engine before investing in figuring out all the nuances of the API & coordinate system. (Note that the API prohibits offline use, as in you can't distribute a processed glTF file like this. But you can do this preprocessing in memory whenever you're fetching tiles). https://ift.tt/610xo4N December 17, 2023 at 10:07PM
Show HN: Modern C++ implementations of a words counter with benchmarks I uploaded a few days ago some code I had around that some people could find interesting, so I share it here. It is a series of increasingly more performant C++20 words counters (though quite simple in the definition of what a "word" is). Feel free to experiment. Things should be ready to use in Linux and Mac. If you find any problems building, please let me know through Github, I do not have much time but when I find a slot I will correct at least user failures. Planning to add a decent `./bootstrap.sh` to speed up dependencies installation by users. https://ift.tt/Ypj4OUT December 17, 2023 at 01:09AM
Show HN: Sqlauthz: Declarative permissions management for PostgreSQL I've been working on a little project recently to solve a problem that I've encountered at every job I've ever had, and I'm eager for some feedback. Having super granular roles & permissions in PostgreSQL is desirable, but quite difficult to maintain (particularly at smaller companies without dedicated security/devops/DBA/whatever who make it their business to maintain it). I've thought for a while that having a declarative way to manage them would be really useful and allow more teams to actually make use of sophisticated security features that PostgreSQL offers in their production systems. You can probably see where this is going... I wrote a tool to do just that! It's called sqlauthz, and it allows you to declarative manage your PostgreSQL permissions by writing rules in the Polar authorization language. https://ift.tt/O2VNuxX December 17, 2023 at 12:42AM
Show HN: Visualising Rising Temperatures for the Netherlands Something I've been working on this weekend: visualising climate change in the Netherlands. https://ift.tt/iX9Lods I particularly liked the heatmap with monthly temperature data since the 1900s. It's a lot of scrolling, but you can clearly see that climate is accelerating since the '90s. https://ift.tt/dGUF45Y December 16, 2023 at 07:19PM
Show HN: Bemi – context-aware data change tracking for Prisma Hey HN, we’re Evgeny, Arjun, and Donna - we’re building automatic data change tracking for PostgreSQL and we’re releasing our Prisma integration! The repo is https://ift.tt/qhOBLfI and our home page is https://bemi.io . We've built robust data tracking systems before, for example at companies that hadn’t implemented an event-sourced architecture from the get-go and needed to scale with traditional transactional databases. We’ve since learned that the tech is useful to lots of others. We’ve gone full-time on Bemi now and have been following that demand to build a general-purpose tracking solution that can be used for a wide range of use cases such as reliable audit trails, surfacing data to customer support, reverting changes made in an API request, or testing different application states. We’ve built Bemi to be lightweight and secure. It takes a practical approach to achieve the benefits of event sourcing without requiring rearchitecting existing code, switching to highly specialized databases, or using unnecessary git-like data versioning abstractions. We want your system to work the way it already does with PostgreSQL to allow keeping things as simple as possible. We plug in at both the database and application levels, to get reliability, performance, and also a comprehensive understanding of every change. On the database level, we ingest changes using change data capture (CDC). On the application level, our library allows passing application context and metadata to the write-ahead logs (WAL) automatically. We then stitch the change data together and store it in a structured format in a destination cloud PostgreSQL. We plan to charge for compute if you’re a company storing >1M changes a month. If this is something that seems interesting to you feel free to play around with it or message us! We’re still early and it’d be amazing to get HN’s feedback. What do you think HN? https://ift.tt/qhOBLfI December 15, 2023 at 11:01PM
Show HN: Screenshot AI, Your Intelligent Screenshot Assistant Few weeks back i have been deleting my screenshots which gave me idea to this side project to declutter and organise my device screenshots. After several weekend it turn out to this. Since deep learning need lot of infra and i am running on budget, results might be slow but you will get results. Let me know in comments what can i do better or how can i do better. https://ift.tt/l4bkFtG December 15, 2023 at 09:53PM
Show HN: I made a long-distance instant camera Howdy! Wanted to share a weekend project of mine. When you take a picture with snapress it'll print on your friends' printers -- kinda like if you separated the camera and printer parts of a Polaroid and put them in different places. Just something to make picture sharing a little more physical again! It's free to use and all you need is a spare computer or raspberry pi to run the script. Everything takes place in the browser, so there's no need to download an app. I made it mostly as a way to connect with family on upcoming travels, it's a nice surprise to come back home to a printed out picture that was printed the moment they took it. Happy to answer any questions about the project! Thanks! https://ift.tt/wMfvN4P December 14, 2023 at 10:17PM
Show HN: FoxColorBox FF exten that allows you to change, customize Window colors FoxColorBox is a Firefox extension that allows you to change and customize browser Window colors. When you open a new window, it will have a distinct color. You can also change colors by clicking on the extension icon. This can help you differentiate between home, work, and school. It can also be useful when working in development, staging and production environments. https://ift.tt/tQj1rNP https://ift.tt/bVmOCgx https://ift.tt/tQj1rNP December 15, 2023 at 12:38AM
Show HN: Octopus – a directed acyclic graph for app development Directed acyclic graphs are muched discussed in comp-sci, but octopus appears to be the first reusable, turnkey, ready-to-wear, off-the-shelf implementation of a DAG for application development, in any language, that I'm aware of. This is remarkable because DAGs hit a sweet spot in the middle of the three common programming paradigms (OO, event-driven, functional). Let's have a DAG as the top-level structure of our applications. Data-fetching and onChange handlers live in DAG nodes, next to the data they act on. The UI flows out from the DAG with fine-grained reactivity. Our app state is effortlessly consistent, because any outside change (user action, api result) unleashes a graph traversal. Our UI components become much simpler, because they just need to dumbly reflect values in the graph. I'm putting this up for a second time. Absolutely no-one bit the first time, which can't be right :-) https://ift.tt/fzVS0FJ December 15, 2023 at 12:06AM
Show HN: Gait Analyzer When ever I visit my spine doctor, the first thing they ask me to do is to walk for checking my gait. I've been wanting to analyze my gait myself for a long time and only now its been possible due to accessible & efficient ML/AI and the ability to run complex models on local computer. Gait abnormalities can be attributed to various musculoskeletal and neurological conditions and so gait analysis is being used as an important diagnostic tool by doctors. Automated gait analysis requires expensive motion capture or multiple-camera systems. But with Gait Analyzer one can analyze their gait in comfort and privacy of their home on their computer. Gait Analyzer implements the algorithm published in the paper titled Automated Gait Analysis Based on a Marker-Free Pose Estimation Model - https://ift.tt/oHxQ5g2 . This algorithm for gait analysis is shown to be as reliable as a motion capture system for most scenarios. Gait Analyzer further uses Llama 2 large language model to interpret the gait data to the end user in simple terms. You can test Gait Analyzer by uploading a short side-view video of you walking. In case you want to run it locally in your computer, It's available on docker and its open-source. I'd appreciate your feedback to improve Gait Analyzer. https://ift.tt/G1PzSw2 https://ift.tt/N3sJrAO December 13, 2023 at 11:35PM
Show HN: Visualize rotating objects from the 4th, 5th, nth dimensions Ever since I remember I had a lot of curiosity regarding hyper dimensional spaces. Picturing higher dimensions, such an impossible yet exciting idea... So years ago I came across a small GIF of a tesseract. Since then it left me wondering how cubes from even higher dimensions would look like... Years passed and I became a software developer, decided to tackle the problem myself and ncube was the result. ncube allows you to visualize rotating hypercubes of arbitrary dimensions. It works by rotating the hyperdimensional vertices and applying a chain of perspective projections to them until the 3rd dimension is reached. Everything is generated in real time just from the dimension number. The application is fully free and open source: https://ift.tt/rRvlS36 . There, you'll find some demos, more detailed explanation and how you can test it out yourself. Binaries for Windows, Mac and Linux are available: https://ift.tt/Vd8JyEa There's also a web version that runs fully on the browser: https://ncube.ndavd.com If you like the project I'd appreciate if you could give it a star on GitHub ♥ If you have any issue or feature request please submit at https://ift.tt/0b3hFfU https://ncube.ndavd.com December 11, 2023 at 11:53PM
Show HN: A dictionary of untranslatable words from around the world Have you ever come across a word in another language that just perfectly captures a feeling or concept you can't quite express in your own tongue? I created coolforeignwords.com because I wanted to share those "aha" moments with fellow language lovers, and created a website to make those words easier to find. So, whether you're a word nerd, a traveler, or just curious about the world, please feel free to visti my website. Try it out and please share your feedback. It's still very early stage, so would love any advice. Thank you https://ift.tt/QsgZhIY December 13, 2023 at 03:04AM
Show HN: QA GPT – Write UI tests in plain English powered by GPT-4-Vision Hey HN, QA GPT enables engineers and QA teams to write UI and functionality tests in plain english. As engineers, we sometimes get a little lazy when it comes to testing the functionality of our changes. It's hard to switch from coder hat to user hat. However, a single bug can significantly impact users experience and satisfaction. Errors found in production aren't just a nuisance; they're costly. The later a bug is discovered, the more expensive it becomes to fix. I built QA GPT as a proof of concept to make writing UI/functionality tests really easy. It's super simple - just write your test case in plain english and run it. For example: - "Test the new sharing functionality by signing in, selecting a user, and clicking share." - "Log in and try adding a product to the cart" - "Create a new card, view the number of the card, and verify the digits match the face of the card" Let me know what you guys think https://ift.tt/KsyCfH4 December 13, 2023 at 02:54AM
Show HN: RΞASON – Open-source TypeScript framework for LLM apps Hi HN! RΞASON is an OSS Typescript framework for developing LLM apps that uses Typescript's interfaces to get structured output from an LLM. While there are other TS LLM frameworks, I think RΞASON fills a unique space in the market: it's laser-focused on only three areas and, most importantly, actively stays away from pre-made prompting & retrieval. I've been in the LLM space since GPT-3 originally came out, and I've always had problems with other frameworks, such as LangChain. I dislike that they focus a ton on out-of-the-box prompting & pre-made agents — I , as the dev, should be the one in charge of it. My belief is that LLMs are a new primitive that programmers can use — not a new way to program; it's still up to the programmer to do the right thing & create the right abstractions. Therefore, it's the developer's job to learn the new concepts that come from this new primitive, such as prompting & retrieval. I see a similar analogy here with ORMs & SQL. What RΞASON helps with is in areas that don't differentiate your app: getting structured outputs, handling streaming, and observability. The goal of RΞASON is to make creating great LLM experiences easier. We try to accomplish this by simplifying the hard stuff & maximizing performance — decreasing as much as possible the TTUB. RΞASON is OpenTelemetry compatible — which allows observability in almost any tool (Zipkin, Jaeger, paid solutions, etc.). I'd really love to hear feedback about RΞASON! It has been a hobby project for the last months and I'm super curious to what y'all will think. By the way, contributions welcome! https://ift.tt/ro52HTV December 13, 2023 at 01:05AM
Show HN: I built an OSS alternative to Azure OpenAI services Hey HN, I am proud to show you guys that I have built an open source alternative to Azure OpenAI services. Azure OpenAI services was born out of companies needing enhanced security and access control for using different GPT models. I want to build an OSS version of Azure OpenAI services that people could self host in their own infrastructure. "How can I track LLM spend per API key?" "Can I create a development OpenAI API key with limited access for Bob?" "Can I see my LLM spend breakdown by models and endpoints?" "Can I create 100 OpenAI API keys that my students could use in a classroom setting?" These are questions that BricksLLM helps you answer. BricksLLM is an API gateway that let you create API keys with rate limit, cost control and ttl that could be used to access all OpenAI and Anthropic endpoints with out of box analytics. When I first started building with OpenAI APIs, I was constantly worried about API keys being comprised since vanilla OpenAI API keys would grant you unlimited access to all of their models. There are stories of people losing thousands of dollars and the existence of a black market for stolen OpenAI API keys. This is why I started building a proxy for ourselves that allows for the creation of API keys with rate limits and cost controls. I built BricksLLM in Go since that was the language I used to build performative ads exchanges that scaled to thousands of requests per second at my previous job. A lot of developer tools in LLM ops are built with Python which I believe might be suboptimal in terms of performance and compute resource efficiency. One of the challenges building this platform is to get accurate token counts for different OpenAI and Anthropic models. LLM providers are not exactly transparent with the way how they count prompt and completion tokens. In addition to user input, OpenAI and Anthropic pad prompt inputs with additional instructions or phrases that contribute to the final token counts. For example, Anthropic's actual completion token consumption is consistently 4 more than the token count of the completion output. The latency of the gateway hovers around 50ms. Half of the latency comes from the tokenizer. If I start utilizing Go routines, might be able to lower the latency of the gateway to 30ms. BricksLLM is not an observability platform, but we do provide integration with Datadog so you can get more insights regarding what is going on inside the proxy. Compared to other tools in the LLMOps space, I believe that BricksLLM has the most comprehensive features when it comes to access control. Let me know what you guys think. https://ift.tt/Yo9RkhU December 12, 2023 at 12:26AM
Show HN: MyYogaFlow – Online Yoga Instructor Hi fellow hackers :) I am an indie hacker based in Switzerland who has been tinkering around with many side projects. Most of the things I developed were for myself or just to get familiar with a specific technology or piece of software. I never really deployed anything so far. Not too long ago, my SO and me bought a deck of cards with different yoga poses on them. Since we both work from our office chairs more than 8 hours a day, we wanted to do something for our body and this is where the yoga idea originated from. After the first 10 minutes of reading through the leaflet and another 30 minutes of choosing the poses and putting them in a somewhat meaningful order, we finally started with our own yoga session. It was clear to me: this is not something my SO wants to do every day. Doing the same order of poses every day wasn't an appealing alternative either. Away with the cards and on the lookout for an alternative again. Hence I started developing an application for us, that automatically generates a yoga session. It shows the poses one by one and guides you through the session. Each session is unique, which makes it more interesting over the days. The app even lets you create your custom flow that you select or it chooses among poses that fit your experience level just right. I have recently launched this application and made it publicly available. It is working (in most of the cases) and we are pretty happy about how it turned out. We used it many times in our lunch breaks or after a stressful day to calm down. I would be more than happy to get some feedback on it and some pointers on what to improve or focus on next. https://ift.tt/TLWnK5X December 10, 2023 at 08:32PM
Show HN: Fine-grained stylistic control of LLMs using model arithmetic We developed a new framework that enables flexible control of generated text in language models. By combining several models and/or system prompts in one mathematical formula, it lets you tweak your style and combine model outputs with ease. A handy tool for those working with LLMs, looking for more fine-grained control of stylistic output. More details in our paper: https://ift.tt/v6uEk7y . Feedback and potential applications are welcome. https://ift.tt/RV0hxdf December 10, 2023 at 02:27AM
Show HN: wallstreetlocal – View investments from the world's biggest SEC filers Hello Hacker News! My name is Anonyo, and I am a seventeen-year-old from Southeast Michigan. This is wallstreetlocal, my passion project for the last year. The United States Securities and Exchange Commission (SEC), forces institutional money managers whose portfolios surpass $100 million to quarterly update the SEC website with their latest investments. These money managers, known as 13F (the form type of the report) filers, are the biggest investors in America, and the law forces them to publicly reveal their investments. The problem I found was that SEC filings are cumbersome to find and view. To remedy this, wallstreetlocal saves and formats SEC data by bulk in an easily accessible format. Other websites do the same thing wallstreetlocal does, but wallstreetlocal places an emphasis on stock data. You can compare prices over time, view gains, and even download the data yourself to make any conclusions you wish. For as many stocks as possible, wallstreetlocal provides accurate, recent data to accompany the filings. As of current, there are 850,000 companies in the search database, 30,000 stocks in the stock database, and about 50 filers queried (equates to a couple million lines of JSON, requested by myself alone). By exploring the website, you can see the resources I used, check out some notable money managers I listed, and download any data that suits you. All for free. I plan to open-source this project someday, but I just wanted to get it released for now. The stack uses FastAPI for the back-end, and Next.js for the front-end. The back-end is hosted on a VM using Docker for the different microservices, and the front-end on Vercel. I made this project to better democratize SEC filings, and also to get some experience on my hands. I love computers, and one day hope to change the world through startups. In the comments, I'd appreciate any and all advice, as well as feedback on how to improve the site. Thanks. https://ift.tt/rBUoSvW December 9, 2023 at 11:26PM
Show HN: Open sourcing Restly – GPT-4 powered tutorials from your OpenAPI spec Hi HN - Marius (mbuleandra) and I built Restly as a side project. Restly generates a user-friendly tutorial against any OpenAPI spec. While building public-facing APIs, we found that writing these guides by hand took a lot of work. Many OpenAPI specs were too large to send to GPT-4 / Claude directly, and even when we could send them, we experienced lost-in-the-middle and hallucination issues. The project is very alpha but works against the Vocode API when we specify well-defined goals. Let us know your thoughts; we would love your contribution to the project! [1] https://getrestly.com/ [2] https://ift.tt/7QjKeWm https://ift.tt/7QjKeWm December 9, 2023 at 04:04AM
Show HN: We've built an Open Source plugin to make using AI easier for VFX Hi, I’m Bryan Lyon, the CTO of DeepMake. We’re making plugins for VFX software suites (Right now we have a pre-release of our After Effects plugin ready) that enable you to use Open Source AI tools without leaving the workflow. It makes all the AI tools work just like another plugin in the VFX software. In the past, I worked on VFX adjacent AI and was more than once asked to make AI tools easier to use so the VFX experts could use them. In many cases my job was little more than making it easy to install and make sense to use. After writing dozens of “one-off” scripts that made an AI tool usable I decided there should be a better way. To that end we’ve simplified the process. Right now our public release is still not the easiest to install, but once you’ve installed one plugin and the backend, installing new plugins is as simple as extracting a zip file or cloning a repo (and we’re going to make it even easier). Our target audience isn’t non-technical people, but rather people who don’t have the time or experience to go through installing an AI tool which may not even include an easy way to send your own data through the pipeline. We hope to make VFX easier for people who can’t really do VFX now due to either a lack of time or experience. We also want to make VFX more powerful for VFX experts. We’ve built the backend to make plugins extremely easy to make and hope that the community will be able to take advantage of an easy way to integrate with VFX tools. IN the spirit of open source we’ve released our code as open source (We’re cleaning up the AE plugin side, but it’ll be released too). We plan on monetizing based on managing massive plugins across compute farms for big VFX shops and other features that benefit those larger teams, the core of the system will remain open source and freely available. So far, we’ve released a Stable Diffusion plugin, and have more including Masking, Super Resolution, and more coming up in the pipeline. I hope you like our software and look forward to hearing from you. https://ift.tt/CdJRq5P December 9, 2023 at 03:15AM
Show HN: Open-Source AI Adventure Game Hi HN! I'd love your feedback for this AI storytelling game. Remember when you used to read choose-your-own adventure books as a kid? We're trying to re-create that. The site lets you create adventures by seeding them with parameters that end up feeding LLMs: title, genre, writing style, etc. Once you've done that, you can play your book like a game: with custom chapters, illustrations, voice narration, choices, and dice rolls. We’re really into the idea of merging the open-ended generative AI with goal-directed story arcs that the game author can control. Everything is open source (links below) and we’d love your contributions & ideas. I know there's a lot of great work in this space, and the whole team that built this is eager to learn. Agent Source Code: https://ift.tt/b7BclnQ Web Source Code: https://ift.tt/VrkcAQO Video playthrough: https://youtu.be/r5iJkELa-kk https://ift.tt/DkQwMNj December 7, 2023 at 10:38PM
Show HN: Biblos – Semantic Search the Church Fathers I'm pleased to present an update to Biblos, a semantic search tool designed for biblical research. This release incorporates instructor-large embeddings to enhance the precision of verse retrieval. Introducing some key features including historical church writings and commentaries alongside the main biblical corpus. Available for use at https://biblos.app/ Technical Insights: - Utilizes Chroma for vector search, now powered by instructor-large embeddings for improved semantic accuracy. - Features Anthropic's Claude LLM model to generate summaries that provide context and clarity for search results. - Developed with a Retrieval Augmented Generation (RAG) architecture, the app offers a streamlined user experience through a Streamlit Web UI, all orchestrated with Python. https://biblos.app/ December 7, 2023 at 11:38PM
Show HN: Online RPG Game with AI being dungeon master Hi I've created a text-based RPG game in which you can experience different adventures and play interesting characters while the AI acts as a real time DM/story teller. There are multiple ways to interact with AI: 1) Input free will move 2) Input default move (affects resources) 3) Post a question to DM Also one more cool feature is that you can create your own adventure via simple form. Add scenes, add characters and make it public so other players are able to test it. I also plan to have 'competetive' side where players will pay some small fee to creators in crypto depending on community, and leaderboards can determine price pool reward distribution. But thats for future. Check it out and tell me what you think guys :) https://v3rpg.com https://ift.tt/FEZWnNQ December 7, 2023 at 01:51AM
Show HN: Lume – automate data mappings using AI Hi HN! I'm Nicolas, co-founder of Lume, a seed-stage startup ( https://www.lume.ai/ ). At Lume, we use AI to automatically transform your source data into any desired target schema in seconds, making onboarding client data or integrating with new systems take seconds rather than days or weeks. In other words, we use AI to automatically map data between any two data schemas, and output the transformed data to you. We are live with customers and are just beginning to open up our product to more prospects. Although we do not have a sandbox yet, here is a video walkthrough of how the product works: https://ift.tt/q5eRcEu?... . And, here is our documentation: https://docs.lume.ai . We would love to get you set up to test it, so please reach out. Using Lume: we do not have self-serve yet. In the meantime, you can request full access to our API through the Request Access button in https://www.lume.ai . The form asks for quick information e.g. email so that I can reach out to you to onboard you. Please mention you came from HN and I’ll prioritize your request. How our full API product offering works: Through Lume’s API, users can specify their source data and target schema. Lume’s engine, which includes AI and rule-based models, creates the desired transformation under the hood by producing the necessary logic, and returns the transformed data in the response. We also support mapper deployment, which allows you to edit and save the AI generated mappers for important production use cases. This allows you to confidently reuse a static and deterministic mapper for your data pipelines. Our clients have three primary use cases - Ingest Client Data: Each client you work with handles data differently. They name, format, and handle their data in their own way, and it means you have to iteratively ingest each new client's data. - Normalize data from unique data systems. To provide your business value, your team needs to connect to various data providers or handle legacy data. Creating pipelines from each one is time consuming, and things as small as column name differences between systems makes it burdensome to get started. - Build and maintain data pipelines. Creating different pipelines to that map to your target schema, whether for BI tooling, downstream data processing, or other purposes, means you have to manually create and maintain these mappings between schemas. We're still trying to figure out pricing so we don't have that on our website yet - sorry, but we wanted to share this even though it's still at an early stage. We’d love your feedback, ideas & questions. Also, feel free to reach out to me directly at nicolas@lume.ai. Thank you. https://www.lume.ai/ December 6, 2023 at 11:07PM
Show HN: Frogtab – Private, peaceful task management Hi HN! I'm excited to share the task manager that I've been building and relying on for the past 9 months I started developing Frogtab after a particularly busy period at work, where a large number of ad-hoc tasks would need my attention each day. What I Initially tried was a single (long!) checklist of outstanding tasks, but I found that too overwhelming to look at all day long. Then I tried scheduling tasks based on when I expected to complete them, but that also required too much cognitive load - I only had the capacity to bucket tasks into "today" and "not today" And so Frogtab was born. Originally as a Google doc, then later as a rudimentary web app with some simple automation. The automation was there to nudge me towards good habits, e.g., grouping all my outstanding tasks at the beginning of each day and requiring me to consciously choose which tasks to tackle that day. I subsequently refined the system a little, but the core mechanic has been serving me well ever since! More than the benefits I've gotten from using Frogtab at work, it's been really fun to develop Frogtab into a form that is usable - dare I say enjoyable - by other people too. Thanks to feedback from the awesome community at Bear (bearblog.dev) and also from sharing an early version here on HN, I've been able to make a whole host of improvements to Frogtab Highlights include: - Support for a weekly routine, aka recurring tasks - Data export and import, including auto-backup in supported browsers - Ability to receive tasks sent from any device - Dark theme - Keyboard shortcuts - Full docs at frogtab.com/help As I've been developing Frogtab, one of my primary concerns has been data privacy. Frogtab stores data in your bowser and doesn't require an account. Also, Frogtab does not use cookies, collect PII, serve ads, or track you in any way. It's entirely free to use and fully open source - the code is at github.com/dwilding/frogtab If you try Frogtab, I hope that you have a great experience! I would love to hear any feedback/suggestions, thank you :) https://frogtab.com/ December 6, 2023 at 04:27AM
Show HN: Solving NYT Connections with ChatGPT Just for fun I decided to see if I could use chatGPT to solve NYT Connections word puzzles. It uses a pretty straightforward BFS search in which the LLM is first prompted to generate several possible groupings of four related words, and then a different prompt is used to evaluate the soundness of each of those groupings. This approach seems to be able to produce the correct solution somewhat less than half the time. Some observations: * For whatever reason, chatGPT-4 seems to be a bit worse than 3.5 at generating Connections groupings. I haven’t tested systematically so maybe this is just some small sample size bias. But at the very least it isn’t obviously better * It really struggles with the “words that can fill in the blank” style groups. Often it will correctly come up with the right category (e.g. “words that can precede `cheese`”) but will only be able to identify 2 of 4 words in that grouping * It frequently generates very vague categories (“words that can be nouns”) despite nothing like that appearing in the proposal prompt. Also it will still sometimes score them highly, despite there being several explicitly examples in the value prompt disallowing these types of categories If you have any idea for how to improve this, please let me know (or send a PR)! https://ift.tt/MoKIdiT December 6, 2023 at 01:41AM
Show HN: Dropbase – Build internal web apps with just Python Hey HN, I’m Jimmy, co-founder of Dropbase ( https://www.dropbase.io ). We are an internal tools builder for Python developers. All you have to do is import any Python scripts/libraries, declare UI components, and layer app permissions so you can share them with others. We’re a middle ground between Airplane and Retool—simpler UI creation than Airplane, more code-centered than Retool. UI building is declarative and you can bind Python scripts/functions to UI components. You can write Python scripts/functions using our App Studio with support from a Python Language Server Protocol (LSP) for linting. Since the self-hosted worker directly references .py or .sql files in the filesystem, you can even write them on VSCode directly or import any other Python script or library. Our app layout is highly opinionated to speed up app building. Instead of an open canvas for UI building, we just give you a main table view and a widget sidebar. This approach significantly reduces app-building time while still covering what most tools need: see some data and take actions based on it. It’s not flexible enough to do absolutely anything, but that’s the point—there’s a tradeoff between flexibility and speed. Dropbase gives you most of what you need, plus speed! A neat feature we are experimenting with to build admin panels fast is “Smart Tables”. We convert any SQL SELECT statement (even across multiple joins and filters) into an inline editable table, like spreadsheets, without any additional code. We have a hybrid hosting model that combines a self-hosted client and a worker server, with a backend API for app/component definitions hosted by us to simplify pushing feature updates. The worker server sits in your machines so your sensitive data doesn’t leave your infra. We’re Python-centric for now, but plan to add support for Rust, Go, and others later. We made a few demo videos building common tools: - Customer approval tool: https://youtu.be/A1MIIRNkv3Q - Data editing tool (with Smart Table): https://youtu.be/R1cHO9lMRXo To try Dropbase, create an account at https://app.dropbase.io and generate a token, then follow these instructions for local setup: https://ift.tt/i39gf5F . We are very early so we're really excited to get your feedback, especially on our approach to tools building with Python! My co-founder Ayazhan and some of our teammates will be around to answer questions. https://ift.tt/KoZdNJ1 December 5, 2023 at 11:55PM
Show HN: Demo of Agent Based Model on GPU with CUDA and OpenGL (Windows/Linux) Demo of agent based model on GPU with CUDA and OpenGL (Windows/Linux) Agent instances on GPU memory Uses SSBO for instanced objects (with GLSL 450 shaders) CUDA OpenGL interops Renders with GLFW3 window manager Dynamic camera views in OpenGL (pan,zoom with mouse) Libraries installed using vcpkg ( https://ift.tt/TMKgHcB ) https://ift.tt/TMKgHcB December 5, 2023 at 01:04AM
Show HN: Satounki – Self-hosted temporary elevated access management service I was reading through the comments on this HN post last night[1] which inspired me to share this project here today. I've been working on Satounki since I got laid off last month. It's the culmination of a lot of experience building similar ad-hoc internal tooling at various places throughout my professional career. Satounki already includes: * AWS support * GCP support * Cloudflare support * Auto-generated Terraform providers from the Rust API * Auto-generated Typescript client wrapper from the Rust API * Slack bot for request notifications, approvals and rejections * CLI for requests, approvals and rejections * Dashboard for exploring policies, requests and stats The scope of this project is pretty big and I'm looking for contributors; if you know Rust, Go, Typescript, Nix, or even have an interest in them, take a look and see if this is something that you might like to work on! [1]: https://ift.tt/MOmJeV6 https://ift.tt/wrFeP0d December 4, 2023 at 10:39PM
Show HN: Audio plugin for circuit-bent MP3 compression sounds I made MAIM, an open-source audio plugin that uses real MP3 encoders to distort the sound. I've also added knobs that let you "circuit bend" the encoders, changing parameters that would normally be inaccessible to the user to get strange glitchy sounds. The plugin lets you switch between two MP3 encoders, since under the MP3 standard, the specifics of what to lose in MP3 lossy compression is left up to the encoder. The encoders are LAME, the gold standard for open-source MP3 encoders, and BladeEnc, an old open-source MP3 encoder that has a really bubbly sound and was fun to work with. I'd love any feedback, and I'll be around to answer questions! https://ift.tt/PkUE4Hg December 3, 2023 at 11:31PM
Show HN: ThreeFold – Decentralized Cloud Infrastructure We built it from the ground up, minimal Linux based operating system (written in go and rust), messaging system written in rust and infrastructure as code support (terraform and pulumi) There is even a dashboard, and a playground UI to manage it (that you can even run yourself) . Beautifully decentralized, you have full control of everything, you can deploy on your own servers if the cost of the cloud is too much. You can even run the whole system yourself if you want to. Here is a link to documentation https://manual.grid.tf/ Also, we are honored if it piqued your interest, and would love to support your testing journey free of charge. https://threefold.io/ December 3, 2023 at 02:13AM
Show HN: AI Shopping Assistant Hi! My friend and I built an AI Shopping Assistant powered by ChatGPT, called ShopMigo. The primary purpose of this project was to learn more about LLMs, embeddings, and AI in general. This started more as a pet project, but turned into something that we actually find really useful. I’m the kind of person that will take weeks to do research before buying an electronic or an expensive product. ShopMigo helps to guide you through a purchase and aggregates reviews so you don’t have to spend weeks/months doing your due diligence. Try it out! Let us know what you think! https://ift.tt/f0kECrH December 3, 2023 at 12:23AM
Show HN: 80% faster, 50% less memory, 0% loss of accuracy Llama finetuning Hi HN! I'm just sharing a project I've been working on during the LLM Efficiency Challenge - you can now finetune Llama with QLoRA 5x faster than Huggingface's original implementation on your own local GPU. Some highlights: 1. Manual autograd engine - hand derived backprop steps. 2. QLoRA / LoRA 80% faster, 50% less memory. 3. All kernels written in OpenAI's Triton language. 4. 0% loss in accuracy - no approximation methods - all exact. 5. No change of hardware necessary. Supports NVIDIA GPUs since 2018+. CUDA 7.5+. 6. Flash Attention support via Xformers. 7. Supports 4bit and 16bit LoRA finetuning. 8. Train Slim Orca fully locally in 260 hours from 1301 hours (5x faster). 9. Open source version trains 5x faster or you can check out Unsloth Pro and Max codepaths for 30x faster training! https://ift.tt/bBRJeA9... has more info about Unsloth! Hopefully you can try it out! Wrote a blog post at https://ift.tt/UmIS4Xw if you want to learn more about our manual hand derived backprop or Triton kernels and stuff! Thanks once again! https://ift.tt/RPDVsIT December 1, 2023 at 08:12PM
Show HN: Block bots and spam on Instagram with SocialGuard Hey HN, I've been working on a project to combat bots and spam on Instagram. SocialGuard provides an overview of all the comments flowing through your IG account, with options to automatically delete or reply to comments based on keywords + phrases. If you use IG for work purposes, I hope it can help keep your posts clean of spam! https://ift.tt/2GKSP7Z December 1, 2023 at 11:11PM
Show HN: Bi-directional sync between Postgres and SQLite Hi HN, Today we’re launching PowerSync, a Postgres<>SQLite bi-directional sync engine that enables an offline-first app architecture. It currently supports Flutter, React Native and web (JavaScript) using Wasm SQLite in the browser, with more client SDKs on the way. Conrad and I (Ralf) have been working on our sync engine since 2009, originally as part of a full-stack app platform. That version of the system is still used in production worldwide and we’ve learnt a lot from its use cases and scaling. About a year ago we started on spinning off PowerSync as a standalone product that is designed to be stack-agnostic. If you’d like to see a simple demo, check out the pebbles widget on the landing page here: https://ift.tt/nlFt6aj We wrote about our architecture and design philosophy here: https://ift.tt/8xaZGLQ... This covers amongst other things how we designed the system for scalable dynamic partial replication, why we use a server authority architecture based on an event log instead of CRDTs for merging changes, and the approach to consistency. Our docs can be found here: https://ift.tt/b4F3iBH We would love to hear your feedback! - Ralf, Conrad, Kobie, Phillip and team https://powersync.com November 30, 2023 at 07:43PM
संदेश
