The best Hacker News stories from Show from the past week
Latest posts:
Show HN: LearnGPT – Browse and share ChatGPT examples
Show HN: LearnGPT – Browse and share ChatGPT examples
Show HN: Web search using a ChatGPT-like model that can cite its sources
We’ve trained a generative AI model to browse the web and answer questions/retrieve code snippets directly. Unlike ChatGPT, it has access to primary sources and is able to cite them when you hover over an answer (click on the text to go to the source being cited). We also show regular Bing results side-by-side with our AI answer.<p>The model is an 11-billion parameter T5-derivative that has been fine-tuned on feedback given on hundreds of thousands of searches done (anonymously) on our platform. Giving the model web access lessens its burden to need to store a snapshot of human knowledge within its parameters. Rather, it knows how to piece together primary sources in a natural and informative way. Using our own model is also an order of magnitude cheaper than relying on GPT.<p>A drawback to aligning models to web results is that they are less inclined to generate complete solutions/answers to questions where good primary sources don’t exist. Answers generated without underlying citable sources can be more creative but are prone to errors. In the future, we will show both types of answers.<p>Examples:<p><a href="https://beta.sayhello.so/search?q=set+cookie+in+fastapi" rel="nofollow">https://beta.sayhello.so/search?q=set+cookie+in+fastapi</a><p><a href="https://beta.sayhello.so/search?q=What+did+Paul+Graham+learn+from+users" rel="nofollow">https://beta.sayhello.so/search?q=What+did+Paul+Graham+learn...</a><p><a href="https://beta.sayhello.so/search?q=How+to+get+command+line+parameters+in+Rust" rel="nofollow">https://beta.sayhello.so/search?q=How+to+get+command+line+pa...</a><p><a href="https://beta.sayhello.so/search?q=why+did+Elon+Musk+buy+twitter" rel="nofollow">https://beta.sayhello.so/search?q=why+did+Elon+Musk+buy+twit...</a><p>Would love to hear your thoughts.
Show HN: Port of OpenAI's Whisper model in C/C++
Hi HN,<p>OpenAI recently released a model for automatic speech recognition called Whisper [0]. I decided to reimplement the inference of the model from scratch using C/C++. To achieve this I implemented a minimalistic tensor library in C and ported the high-level architecture of the model in C++. The entire code is less than 8000 lines of code and is contained in just 2 source files without any third-party dependencies. The Github project is here:<p><a href="https://github.com/ggerganov/whisper.cpp" rel="nofollow">https://github.com/ggerganov/whisper.cpp</a><p>With this implementation I can very easily build and run the model - <i>“make base.en”</i>. It also allows me to run it on a wide range of devices. For example, I have provided examples of running the model on an iPhone, Raspberry Pi 4 and even in a web page via WebAssembly!<p>The implementation runs fully on the CPU and utilizes FP16, AVX intrinsics on x86 architectures and NEON + Accelerate framework on Apple Silicon. The latter is especially efficient and I observe that the inference is about 2-3 times faster compared to the current PyTorch implementation provided by OpenAI when running it on my MacBook M1 Pro. The WASM port utilizes SIMD 128-bit intrinsics - a feature supported in some modern web browsers [1].<p>I am very happy with the performance that I observe on Apple Silicon devices. I didn’t expect that the Accelerate framework [2] (i.e. CBLAS) offers such a dramatic performance boost for matrix multiplications so I was very pleasantly surprised! To enable the framework in your C/C++ projects, all you have to do is add <i>`-framework Accelerate`</i> to your clang command-line flags.<p>This entire exercise of implementing the Whisper model was very interesting to me and helped me understand a lot about how the transformer architecture works. I also got a lot of positive feedback from people finding and using my project. We brainstormed on a lot of interesting tools that can potentially be created with this library (such as speech-to-text plugin for Vim, RPi4 voice assistant, WASM chat bot, etc). If interested, checkout the “Examples” section and the “Show and tell” discussions for some ideas!<p>Would love to know what you think about this project and about your experience with using the Accelerate framework in any of your projects.
Cheers!<p>[0] <a href="https://github.com/openai/whisper" rel="nofollow">https://github.com/openai/whisper</a><p>[1] <a href="https://chromestatus.com/feature/6533147810332672" rel="nofollow">https://chromestatus.com/feature/6533147810332672</a><p>[2] <a href="https://developer.apple.com/documentation/accelerate" rel="nofollow">https://developer.apple.com/documentation/accelerate</a>
Show HN: Domain Name Search with AI
In my exploration of OpenAI, I just created a domain-name search that takes business description as an input, and generates interesting domain names for it. It then uses DNSimple API to check if .com is available.<p>In my view it is a much easier way to find a suitable domain, as the AI thinks of a much large pool of possible names than my own brain. SmartyNames found its own name, using the tool itself.<p>Hope you enjoy it! <a href="https://smartynames.com/" rel="nofollow">https://smartynames.com/</a>
Tell HN: My child's first program
Last night, I introduced my kid to programming. We'd done some stuff with Mindstorms before, but she never really caught the bug for it. But for some reason, last night when I showed her some simple Python scripting to solve math problems and write to the console, she was enthralled.<p>After guiding her through a few things, she took the laptop off for a while and then came back with her first program, giggling like a maniac<p><pre><code> you='WOW!!!'
fart='So many poops!'
print(you,fart)
</code></pre>
I'm pretty proud :D
Tell HN: Giving ChatGPT access to a real terminal
So, I guess this is the inevitable conclusion with LLMs. Connect them to a real terminal and let them act on real-world objects... I honestly don't know whether I like the idea or not, but I guess it's good to have this conversation now while it is only a marginally better version of tldr.<p>But you can already use it do do simple tasks like cleaning old files, figuring out what machine you're running on or even perform and summarize portscan results.<p>It should go without saying that this should be done on VMs and every command is confirmed and checked by the user...<p>tldr: browsing: enabled
Show HN: Whole Git repo was made with ChatGPT
Show HN: Whole Git repo was made with ChatGPT
Show HN: Chrome extension to display ChatGPT response besides Google Search
Show HN: Chrome extension to display ChatGPT response besides Google Search
Show HN: Explore Wikipedia edits made by institutions, companies and governments
Hi HN!<p>Wikiwho is a tool that scans Wikipedia edits and extracts those coming from specific IP ranges associated to known organizations. I've made this as a for-fun side project two years ago.<p>If you want to read more on how it works I've written a short blog article about it here: <a href="https://ailef.tech/2020/04/18/discovering-wikipedia-edits-made-by-institutions-companies-and-government-agencies/" rel="nofollow">https://ailef.tech/2020/04/18/discovering-wikipedia-edits-ma...</a><p>I had already posted it here at the time (previous discussion: <a href="https://news.ycombinator.com/item?id=22907200" rel="nofollow">https://news.ycombinator.com/item?id=22907200</a>) but I've now decided to release the code openly, hence the repost.<p>If you're insterested, you can check the repo here: <a href="https://github.com/aileftech/wikiwho" rel="nofollow">https://github.com/aileftech/wikiwho</a> (disclaimer: the code is a bit clumsy).<p>Cheers!
Show HN: This Hacker News Does Not Exist
Show HN: I wrote a free eBook about many lesser-known/secret database tricks
Show HN: I wrote a free eBook about many lesser-known/secret database tricks
Show HN: Natural Language Processing Demystified
Link: <a href="https://www.nlpdemystified.org/" rel="nofollow">https://www.nlpdemystified.org/</a><p>Hi HN:<p>After a year of work, I've published my free NLP course. The course helps anyone who knows Python and a bit of math go from the basics to today's mainstream models and frameworks.<p>I strive to balance theory and practice, so every module consists of detailed explanations and slides along with a Colab notebook putting the ideas into practice (in most modules).<p>The notebooks cover how to accomplish everyday NLP tasks including extracting key information, document search, text similarity, text classification, finding topics in documents, summarization, translation, generating text, and question answering.<p>The course is divided into two parts. In part one, we cover text preprocessing, how to turn text into numbers, and multiple ways to classify and search text using "classical" approaches. And along the way, we'll pick up valuable bits on how to use tools such as spaCy and scikit-learn.<p>In part two, we dive into deep learning for NLP. We start with neural network fundamentals and go through embeddings and sequence models until we arrive at transformers and the mainstream models of today.<p>No registration required: <a href="https://www.nlpdemystified.org/" rel="nofollow">https://www.nlpdemystified.org/</a>
Show HN: SinglePage – Quickly and anonymously publish a page to the web
Creating a basic webpage has become way too complicated and expensive. Often there are those times when you just want to share your thoughts with the world but don't want the overhead and complexities that come with maintaining a website. Sometimes, you have an interesting thought piece, an education article, or just a quick and simple bio page that doesn't need the heavy hand of a WordPress blog or Medium post. That's where Single Page comes in. Publish a single page instantly to the web with no fuss.<p>I was laid off three weeks ago from Twitter and I decided to work through a couple of my projects and this was one of them. I've tried blogs over the years, Medium didn't feel right but yet I wanted to quickly post pages online and couldn't find an easy way to do it. So I created it.<p>Feedback appreciated!
Show HN: SinglePage – Quickly and anonymously publish a page to the web
Creating a basic webpage has become way too complicated and expensive. Often there are those times when you just want to share your thoughts with the world but don't want the overhead and complexities that come with maintaining a website. Sometimes, you have an interesting thought piece, an education article, or just a quick and simple bio page that doesn't need the heavy hand of a WordPress blog or Medium post. That's where Single Page comes in. Publish a single page instantly to the web with no fuss.<p>I was laid off three weeks ago from Twitter and I decided to work through a couple of my projects and this was one of them. I've tried blogs over the years, Medium didn't feel right but yet I wanted to quickly post pages online and couldn't find an easy way to do it. So I created it.<p>Feedback appreciated!
Show HN: Trading cards made with e-ink displays
I made a thing!<p>In 2014, I was holding a stack of iPhones and thought to myself:<p><pre><code> "Hey, if I had each phone display a playing card, I could click a button and they'd shuffle themselves"
</code></pre>
I pared that idea all the way down to this: trading cards made of e-ink displays.<p>Right now, each card costs me about $20 each, but with only a bit more scale, I think I can get that down to $10.<p>In doing this project, I learned how to design electronics and circuit boards. I learned Rust and wrote my first driver, I upped my CAD skills, 3D printed, and did my first resin casting. I generated the images on the cards using stable-diffusion.<p>HN always seems to appreciate new uses for e-ink. Thought I'd share :)
Show HN: Trading cards made with e-ink displays
I made a thing!<p>In 2014, I was holding a stack of iPhones and thought to myself:<p><pre><code> "Hey, if I had each phone display a playing card, I could click a button and they'd shuffle themselves"
</code></pre>
I pared that idea all the way down to this: trading cards made of e-ink displays.<p>Right now, each card costs me about $20 each, but with only a bit more scale, I think I can get that down to $10.<p>In doing this project, I learned how to design electronics and circuit boards. I learned Rust and wrote my first driver, I upped my CAD skills, 3D printed, and did my first resin casting. I generated the images on the cards using stable-diffusion.<p>HN always seems to appreciate new uses for e-ink. Thought I'd share :)