Show HN: Vapi – Convince our voice AI to give you the secret code

My wife and I just launched this app because we had so much fun with a fitness challenge we did we some friends recently where everyone put in $100 and defined their goals, then only the people who met their goals split the pot. It made us realize there&#x27;s really no better motivator than money.It&#x27;s better on mobile screens right now as we just designed for mobile screens for now. And we did manage to put it up on both app stores - it&#x27;s low in the ranks but you should be able to search it by &quot;Goalie challenges&quot;.We built it using the leanest tech stack - Ruby on Rails, server-side HTML (no JS), tailwindcss, then packaging it for iOS and Android using pwabuilder!

Show HN: An app that takes your money if you don't reach your goals

Here&#x27;s a project I&#x27;ve been working on for the last few months.It&#x27;s a new (I think) algorithm, that allows to adjust smoothly - and in real time - how many calculations you&#x27;d like to do during inference of an LLM model.It seems that it&#x27;s possible to do just 20-25% of weight multiplications instead of all of them, and still get good inference results.I implemented it to run on M1&#x2F;M2&#x2F;M3 GPU. The mmul approximation itself can be pushed to run 2x fast before the quality of output collapses.The inference speed is just a bit faster than Llama.cpp&#x27;s, because the rest of implementation could be better, but with a better development I think it can be a new method to speed up inference - in addition to quantization.You could call it ad-hoc model distillation :)You can change the speed &#x2F; accuracy of a model at will, in real time.Oh, and as a side effect, the data format allows to also choose how much of the model you want to load into the memory. You can decide to skip say 10-20-40% of the least important weights.It&#x27;s implemented for Mistral, it was also tested slightly on Mixtral and Llama. It&#x27;s for FP16 for now, but Q8 is in the works.The algorithm is described here, and the implementation is open source.<a href="https:&#x2F;&#x2F;kolinko.github.io&#x2F;effort&#x2F;" rel="nofollow">https:&#x2F;&#x2F;kolinko.github.io&#x2F;effort&#x2F;</a>I know these are bold claims, but I hope they survive the scrutiny :)

Show HN: Speeding up LLM inference 2x times (possibly)

Hello HN,in 2017 I started a project that would become MonsterWriter. First envisioned as a semi-structured wiki it became a writing application specialized for scientific content. It is a perfect tool if you write your thesis. While it is focused on technical content, you can still see the knowledge management spirit in it.One or two years ago, my wife joined me in my efforts and redesigned the whole project and we recently released it as MonsterWriter2.To celebrate this milestone we are giving away free lifetime licenses for the Desktop version (till Apr 28). Just use the promo code &quot;gu0ho4q&quot; for a 100% discount. You can find detailed instructions here: <a href="https:&#x2F;&#x2F;www.monsterwriter.app&#x2F;promotion.html" rel="nofollow">https:&#x2F;&#x2F;www.monsterwriter.app&#x2F;promotion.html</a>You can also find a short introduction video to the app on YouTube: <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=vR8i-EY_UBk" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=vR8i-EY_UBk</a>We are happy to receive any kind of feedback!

Show HN: MonsterWriter – Write a thesis, post, or organize notes

Hey everyone!I’m excited to share a project I’ve been working on throughout my university studies. It’s called BiTE (<a href="https:&#x2F;&#x2F;github.com&#x2F;WINSDK&#x2F;bite">https:&#x2F;&#x2F;github.com&#x2F;WINSDK&#x2F;bite</a>) and it&#x27;s a tool primarily focused on being an executable viewer with reverse engineering capabilities.BiTE supports Windows, MacOS, and Linux, along with their associated executable formats. It’s also capable of parsing and displaying debug information using DWARF&#x2F;PDB formats, which I hope will be useful even for just comparing codegen.I’ve put a lot of effort into this and it&#x27;s the first time I&#x27;m releasing something like this publicly. Any feedback, bug reports, or feature suggestions would be greatly appreciated!

Show HN: BiTE – Cross-platform executable viewer and reverse engineering tool

Hi HN – Excited to share a beta for Minard, a new data visualization toolkit we&#x27;ve been working on that lets you generate publication-quality charts with simple natural language (throw away your matplotlib docs and rejoice!).Upload or import CSVs, Excel, and JSON, give it a spin, and please let us know what you think! (Long format data works best for now)For those curious, the stack is a simple Django app with HTMX&#x2F;Alpine and all of the charts are specified and rendered as Vega (<a href="https:&#x2F;&#x2F;vega.github.io&#x2F;vega&#x2F;" rel="nofollow">https:&#x2F;&#x2F;vega.github.io&#x2F;vega&#x2F;</a>). Lots of LLM function calling under the hood as well.

Show HN: Minard – Generate beautiful charts with natural language

Hello HN, I&#x27;m Fares A. Bakhit a junior CS student at Cairo University and I&#x27;m happy to announce my latest project,&quot;Ayin&quot; is an open-source photo editing software available on Windows, Linux, and MacOS (Only a Windows build is available now on GitHub but you can compile it yourself to other platforms) with an interactive real-time graphical user interface.Feel free to use and study the source code of Ayin, it&#x27;s available on GitHub under the GPLv3 open-source license.I&#x27;ve made this project in part of the Winged Dragon Competition at the Faculty of Computers and Artificial Intelligence (FCAI-CU) Cairo University.I&#x27;ve used ImGui, SDL2, and OpenGL (for image rendering, not processing)A showcase video in Arabic is available, showing the functionalities of Ayin: <a href="https:&#x2F;&#x2F;youtu.be&#x2F;ogkteQkJb0I" rel="nofollow">https:&#x2F;&#x2F;youtu.be&#x2F;ogkteQkJb0I</a>Any feedback is much appreciated, thank you!

Show HN: Ayin – An open-source photo editing software

Show HN: Empaithy – AI Microjournaling for Mental Wellness and Habit Building

This took me 3 years to finish. (It is 100% self-published, not endorsed by Google.)So… I wrote a book. It’s a different book with a unique approach. It’s not a novel or a technical book. It’s a biography, a company’s biography. My hope is that it serves two purposes: to inspire founders and to captivate interior designers.It all started three years ago. My wife and I were discussing interior design for our living room and she brought up what beautiful coffee table books should we have. After looking at her favorites I realized there’s nothing like this for tech startups. Digital products are intangible, how cool would it be to make them tangible and a beautiful art&#x2F;decor book for your home office or living room?This idea of consolidating my particular interests with my wife’s decorative goals got stuck in the back of my head for a while. I searched and searched. I could only find these kinds of books related to design, architecture, fashion, or travel. Nothing about tech startups. The next logical step for me was to do it myself…And oh… what a “mistake” it was. This whole project was done during nights and weekends - the first half while working on my startup and the last half while helping create Webflow Labs. You have no idea how many times I “quit” the project. In the end, perseverance and discipline were key to making it happen.During my founder journey, I collected several notes about successful (and unsuccessful) tech startups to help me learn from them. I decided to write about Google’s fascinating story: a true generational company and one of the most valuable companies in the world.It’s with great pleasure and satisfaction that I’m introducing you to “Google - First 25 Years”. A book celebrating the minds behind the tech giant&#x27;s incredible journey from Stanford to global dominance.I have absolutely no idea if this book’s format and concept make sense or if it has an audience willing to invest ($169). Here are some of the unique features:▪ Silk hardcover▪ Limited edition▪ Handpainted edges▪ More than 30 handmade illustrationsI’m producing only 1,000 copies. If the book concept seems interesting enough, leave a comment, I’d love to know your thoughts!Either way, this project was a super fun ride where being outside my comfort zone was the best way to learn storytelling, graphic design, and product manufacturing.At least, I hope this inspires you to go back to your side projects and finish them!

Show HN: A self-published art book about Google's first 25 years

Hi HN! Like many of you, for my entire career I have relied on Homebrew to install all kinds of software on my Mac. That&#x27;s why today I&#x27;m really excited to share a new app that my partner and I are building: Brewer X, a refreshing user interface for Homebrew.Brewer X is graphical interface that lives on top of Homebrew. Leveraging the power of native APIs improves the classic experience and unlocks new features. For example:• bulk actions are performed in parallel• syncing the entire library locally provides incredible search performances and the ability to query descriptions and other previously unaccessible fields• maintenance scripts run automatically for you• last but not least... app icons (or favicons when not available) let you quickly identify what you&#x27;re looking forThe app is written in Swift and uses only AppKit with Nib files for top performance, pixel perfect design, and maximum flexibility.We also designed the app icon and all the others in the UI ourselves. Following the great insights from Sketch[^1] we managed to make them super crispy. We&#x27;re also very proud to have been featured in the macOS App Icon Book[^2].The app has only been out for a couple of weeks, but we&#x27;ve already seen an amazing response from the community. We can&#x27;t tell you all of our future plans yet, but here&#x27;s a list of things we&#x27;d like to see in the app in the near future:• Import&#x2F;Export• Automatic replacement of apps installed without Homebrew• Notifications about available updates• Finder actions• Spotlight integrationI hope you find Brewer X interesting. We&#x27;re happy to answer any question![^1]: <a href="https:&#x2F;&#x2F;www.sketch.com&#x2F;blog&#x2F;how-we-redesigned-our-toolbar-icons-for-big-sur-and-monterey&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.sketch.com&#x2F;blog&#x2F;how-we-redesigned-our-toolbar-ic...</a>[^2]: <a href="https:&#x2F;&#x2F;www.appiconbook.com&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.appiconbook.com&#x2F;</a>

Show HN: Brewer X, a native macOS client for Homebrew

Show HN: A minimal workflow orchestrator for Python written in Rust

Hey HN! I&#x27;m from Brazil and I created Term Typer to help my little brother learn other languages while practicing his keyboard typing skills. We&#x27;ve found it super helpful and fun. Feel free to try it out and let me know your thoughts and feedback. Thanks a lot!

Show HN: Term Typer – Learn a language by typing

Balter, A Load TestER, is a new distributed load testing framework I&#x27;ve been working on for the past few months. It is Open Source, and designed from the ground up to support composability. This allows flexibility in the kinds of load tests you are able to run, such as fully stressing the system and then automatically turning off services; or the ability to run a background noise test, and then slam the service with traffic.Let me know what you think!

Show HN: Balter, the Open Source, composable, distributed load testing framework

Show HN: Cards Against Agility, a fun game for agile fans and cynics alike

Show HN: How to program an Arduino using Rust (on macOS)

For the last few weeks, I have been working on training an LLM from scratch and deploying it in production on Google Cloud Platform.
Finally, I trained a 30 million parameter model on 1 billion tokens and deployed it as a web service.
You can access the LLM using this site - <a href="https:&#x2F;&#x2F;kunalmishra.info" rel="nofollow">https:&#x2F;&#x2F;kunalmishra.info</a>The following steps were taken to build Tiny Tales GPT
1. Downloaded and preprocessed 8GB of dataset using multiprocessing library.
2. Tokenized the data using byte pair encoding to create 1 billion tokens sharded in different bin files.
3. Defined a training setup and trained the model on a small version of the LLaMA model architecture with 30 million parameters.
4. The training was done using Distributed Data-Parallel on two A-100 GPUs provided by JarvisLabs.ai (they are most cost-optimized)
5. After the training is done, an inference script is created to predict the tokens from the trained model given the input context vector.
6. Developed REST-based API service using Flask framework to interact with the inference service to the end user.
7. Finally used GCP&#x27;s virtual machines, instance groups, load balancers, and DNS services to deploy the service on the internet.

Show HN: Trained Tiny Tales GPT(30M model)from scratch and deployed in $15

Hi! We are excited to announce the second release of Desbordante — an open-source, high-performance data profiler that is capable of discovering and validating many different patterns in data using various algorithms.Unlike existing data profilers, Desbordante focuses on discovering complex patterns in data, which are notoriously hard to extract. Since its inception in 2019, it has become the fastest open-source tool for these tasks. It also offers an array of patterns which have no alternative implementations. With this release, Desbordante now supports 17 types of patterns, such as: various types of functional dependencies, inclusion and order dependencies, fuzzy algebraic constraints and many others.Some ways in which Desbordante can be helpful are:
1) Hypothesis generation for scientists that work with large volumes of data.
2) Business data owners and business analysts can benefit from hypothesis generation as well as data quality improvement: cleaning databases from errors, finding and removing inexact duplicates, and so on.
3) Found primitives can help data scientists in feature engineering and choosing the right direction for ablation studies.Desbordante solves two types of tasks: Discovery and Validation. The Discovery task is designed to identify all instances of a specified pattern type of a given dataset. The Validation task is different: it is designed to check whether a specified pattern instance is present in a given dataset. This task not only returns True or False, but it also explains why the instance does not hold (e.g. it can list table rows with conflicting values).Desbordante offers a CLI, a web application, and a Python library. The latter makes it possible to construct ad-hoc data analysis pipelines — essentially, your own applications for various data quality tasks: data cleaning, data deduplication, anomaly detection, data schema recovery and many others. You can check out example implementations here: <a href="https:&#x2F;&#x2F;github.com&#x2F;Desbordante&#x2F;desbordante-core&#x2F;tree&#x2F;main&#x2F;examples">https:&#x2F;&#x2F;github.com&#x2F;Desbordante&#x2F;desbordante-core&#x2F;tree&#x2F;main&#x2F;ex...</a>.Check out some of our articles for more details:<a href="https:&#x2F;&#x2F;medium.com&#x2F;@chernishev&#x2F;exploratory-data-analysis-with-desbordante-4b97299cce07" rel="nofollow">https:&#x2F;&#x2F;medium.com&#x2F;@chernishev&#x2F;exploratory-data-analysis-wit...</a><a href="https:&#x2F;&#x2F;itnext.io&#x2F;building-a-simple-data-cleaning-application-with-desbordante-e4897dcd4c5d" rel="nofollow">https:&#x2F;&#x2F;itnext.io&#x2F;building-a-simple-data-cleaning-applicatio...</a><a href="https:&#x2F;&#x2F;levelup.gitconnected.com&#x2F;checking-mining-and-exploring-functional-dependencies-in-python-903bb0e26d5d" rel="nofollow">https:&#x2F;&#x2F;levelup.gitconnected.com&#x2F;checking-mining-and-explori...</a>This major release brings a lot of improvements: support for several novel patterns, support for novel data type — graphs, added python bindings for existing patterns, better guides and examples and more. The detailed changelog can be seen here (<a href="https:&#x2F;&#x2F;github.com&#x2F;Desbordante&#x2F;desbordante-core&#x2F;releases&#x2F;tag&#x2F;v2.0.0">https:&#x2F;&#x2F;github.com&#x2F;Desbordante&#x2F;desbordante-core&#x2F;releases&#x2F;tag...</a>).

Show HN: Desbordante 2.0 – A high-performance data profiler

Dimidium is a standard-looking terminal color scheme. All colors have been adjusted to have fine visibility.Crafted with Color Appearance Model, it considers the perceptual aspects(lightness, hue) of human color vision.It aims to be a decent default for terminals.Preview, Download: <a href="https:&#x2F;&#x2F;github.com&#x2F;dofuuz&#x2F;dimidium">https:&#x2F;&#x2F;github.com&#x2F;dofuuz&#x2F;dimidium</a>Detailed explaination how I made the color scheme:
<a href="https:&#x2F;&#x2F;dofuuz.github.io&#x2F;color&#x2F;2024&#x2F;03&#x2F;17&#x2F;dimidium-terminal-color-scheme.html" rel="nofollow">https:&#x2F;&#x2F;dofuuz.github.io&#x2F;color&#x2F;2024&#x2F;03&#x2F;17&#x2F;dimidium-terminal-...</a>The code used to make color scheme is also present in the repository.

Show HN: Dimidium – Crafting Terminal color scheme with color science

Show HN: I made an AI fortune teller to predict the future

Show HN: I Made a ROS Package for Realtime Semantic Segmentation

Hi HN,Over the past week, I&#x27;ve been developing Cursumber, a tool designed to help you easily create and integrate unique cursors into your website. Today marks the soft launch of this platform, and I&#x27;m eager to gather community feedback and gauge market interest.Features of Cursumber:Upload and Customize: Select or upload your cursor images (supports PNG, JPG).Preview in Real-Time: Test how your cursor looks and functions on your site before going live.Easy Integration: Simply copy the provided JavaScript script and paste it into your website&#x27;s code.Community Building: (Planned) Future plans include creating a community space where users can share and discuss their custom cursor designs.Cursumber aims to add a layer of identity and uniqueness to your projects with minimal effort. I&#x27;ve put a lot of work into making this tool user-friendly and would really appreciate your feedback and suggestions.Thanks for checking it out—I&#x27;m looking forward to hearing what you think!

Show HN: I made a custom cursor builder in 1 week

While building mendable - we found that feeding LLMs well-structured markdown improved accuracy. We also found it surprisingly hard.We found some great tools online, but none reliably handled the entire process. We wanted an API that took a URL, crawled the pages in the URL, and gave us an easy-to-use, up-to-date markdown we could feed into our index.So, we released an open-source repo and an API that crawls and turns entire websites into a markdown with just a few lines of codeThe API handles:- Crawling without consistent sitemaps
- Infra to handle running many crawling jobs
- Proxying, hosting headless browsers at scale
- Conversion to clean markdown
- Caching
- Handling images, videos (soon), and tables(soon)
- LLM extraction (soon)It is open source, and we also offer an easy-to-use API that starts free. It has built-in loaders for both @llama_index and @langchain.Excited to see people try it

Show HN: API to turn entire websites into Markdown

Google’s recent announcement of a proposed ‘Device Bound Session Credentials’ feature[1] for Chrome reminded me of a project we worked on last year at my company. We focus on fraud prevention at signup and login (preventing multi-accounting and account theft), but some customers were concerned about post-login security and asked us to add a session hijacking prevention feature to our fraud prevention API. In the end, we decided to just implement a solution in Javascript. We call it session-lock, and it can be used today across all browsers[2] and, theoretically, native mobile apps.For a more comprehensive writeup and quick demo, you can visit the link. In short, the library adds a device-bound signature to the session token each time it’s used. At login, it creates a signing key pair on the browser using SubtleCrypto, with the private key set as “unextractable” and stored in IndexedDB (this forms the basis of its device-binding). Also at login, the public key is sent to the server along with the user’s credentials. If the credentials are valid, the server adds the public key to the payload of the JWT it returns to the client. When the client uses the JWT to access a protected resource, session-lock adds a signature to the end of it, along with a timestamp to mitigate replay attack risk. The server then validates the signature using the public key embedded in the JWT. The private key must be removed from IndexedDB upon logout.While Chrome’s DBSC would be a nice solution when it gets around to being deployed in Chrome and Edge, I think session-lock would help a lot today across all browsers in countering common attacks like malicious browser extensions that rip LocalStorage and cookies. Also, implementing the general flow in native mobile apps[3] would have the same key advantages as DBSC - compiled code already on the device and utilization of hardware TPMs.Aside from JS tampering and extracting “unextractable” CryptoKeys from IndexedDB, please let me know if you can think of any other potential attacks. Happy to answer any questions.[1] <a href="https:&#x2F;&#x2F;blog.chromium.org&#x2F;2024&#x2F;04&#x2F;fighting-cookie-theft-using-device.html" rel="nofollow">https:&#x2F;&#x2F;blog.chromium.org&#x2F;2024&#x2F;04&#x2F;fighting-cookie-theft-usin...</a>
[2] Other than Firefox private browsing mode due to its blocking of IndexedDB
[3] Using CryptoKit &#x2F; KeyStore for ECDSA

Show HN: Device-Bound Session Tokens in JavaScript

Show HN: Bridge – Automatic Infrastructure for Django

Show HN: A new open-source automation tool as an alternative to Ansible/Salt

Homepage: <a href="https:&#x2F;&#x2F;loki.librai.tech&#x2F;" rel="nofollow">https:&#x2F;&#x2F;loki.librai.tech&#x2F;</a>
Github: <a href="https:&#x2F;&#x2F;github.com&#x2F;Libr-AI&#x2F;OpenFactVerification">https:&#x2F;&#x2F;github.com&#x2F;Libr-AI&#x2F;OpenFactVerification</a>

Show HN: Loki Needs You – Collaborate on an Open-Source Fact-Checking AI

Termax leverages the power of LLMs and RAG to dynamically generate shell commands, simplifying complex tasks and enhancing productivity. 1. converts your natural language to commands 2. guess next command 3. ease of use plugin (Ctrl-K)Thanks for the feedback from the Hacker News Community. We are aware of developers&#x27; concerns regarding data safety in the application of LLM and RAG during development. We have enhanced data safety in our project and will keep doing so in the future. Please read the tip and note written in the repo readme before using any of our features to protect your data.

Show HN: A shell CLI tool powered by LLM and RAG (include data security tips)

Download for free:
https:&#x2F;&#x2F;chromewebstore.google.com&#x2F;detail&#x2F;easy-folders-chatgpt-clau&#x2F;gdocioajfidpnaejbgmbnkflgmppibfe?utm_source=hackernewsA simple and powerful tool to organize chats in ChatGPT and Claude.** NEW! Claude support added - Create folders and organize your chats for Claude **
** NEW! Multi-account support added - Create folders and organize your chats for multiple ChatGPT &amp; Claude accounts, and switch between them seamlessly. **- Privacy focused, your data is all stored on your local machine 
- Create folders &amp; sub-folders for your chats 
- Easy-to-use drag &amp; drop 
- Colored folders 
- Automatic folders for Custom GPTs 
- Sync folders across multiple devices 
- Full chat history search 
- Bookmark important chats 
- Bulk delete chats 
- Backup &amp; Import data 
- Dark &amp; light mode support

Show HN: Easy Folders – A folders based organization tool for ChatGPT and Claude

Hi HN! I&#x27;m here to share Vibebox (vibebox.app), a little toy app I built to sprinkle joy across your inbox with friendly animal avatars and messages that transform mundane emails into warm, joyful reads.About Vibebox- Transforms your emails with playful visuals and warm messages.- Aims to reduce stress associated with email management.- Promotes positivity within your daily routine.I&#x27;d love your feedback!Is this something you&#x27;d actually use if it were built out? I&#x27;m convinced it&#x27;s a fun toy app, but a few early beta testers have said it&#x27;s turning into their favorite way to read email because it actually reduces the emotional impact of opening their inbox.Inspiration: 
As you can imagine, Vibebox uses LLMs behind the scenes to vibe-ify your emails. If you&#x27;re curious about how it works, I gave a talk on an early iteration of Vibebox called &quot;AI for mere mortal web developers&quot; at Austin API Summit last month.- Video: <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=pTQI0emPLDM" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=pTQI0emPLDM</a>- Repo: <a href="https:&#x2F;&#x2F;github.com&#x2F;nylas-samples&#x2F;austin-api-summit-2024">https:&#x2F;&#x2F;github.com&#x2F;nylas-samples&#x2F;austin-api-summit-2024</a>

Hacker News | Show