The best Hacker News stories from Show from the past day
Latest posts:
My Failure Resume
Show HN: Alpaca.cpp – Run an Instruction-Tuned Chat-Style LLM on a MacBook
Show HN: Alpaca.cpp – Run an Instruction-Tuned Chat-Style LLM on a MacBook
Show HN: Alpaca.cpp – Run an Instruction-Tuned Chat-Style LLM on a MacBook
Show HN: Alpaca.cpp – Run an Instruction-Tuned Chat-Style LLM on a MacBook
Show HN: Alpaca.cpp – Run an Instruction-Tuned Chat-Style LLM on a MacBook
Show HN: Open source UI for ChatGPT API
<a href="https://github.com/KTruong008/aichatbestie">https://github.com/KTruong008/aichatbestie</a><p>Been working on this since last week and glad to share it here and now!<p>There's already been a few other takes on this idea, but I figured none would fit me as well as if I had just built it myself.<p>All messages are sent and stored locally in the browser for privacy and speed. Accounts are optional and upgrading is a one-time deal.<p>Any questions, comments, feedback, please, I'd like to hear it all!
Show HN: Write Language-agnostic functions with WebAssembly you can use
Hey y’all, we’re really excited to share that we’re launching a new free, open source tool called Scale. It’s a highly-performant WebAssembly function runtime that enables composable, language-agnostic software development.<p>TL;DR - Scale allows you to write functions in any language, then use them in any other language. It’s pretty exciting.<p>We currently support Go and Rust, but TypeScript is coming soon and support for a handful of other languages is already in the works. We’d love to hear any feedback you have, so let us know what you think!
Show HN: Mr. Graph – A graph definition and execution library for Python
What:<p>Mr. Graph is a python library designed to make composing graphs of sync and async functions easy! Use google style docstrings to automagically create dataclasses and chain together function calls into graphs.<p>Why:<p>I like the design of Dagster, but not the latency. For apps and systems engineering, sometimes I want to compose a graph out of regular python functions. I don;t need all the heavy machinery that comes with a full workflow engine.<p>Current features:<p>- Use with either async or sync functions<p>- Uses google style doc strings to name return values.<p>- Creates dataclasses for each function's output.<p>- Can infer pipelines from input and output signatures<p>- All directed acyclic graph layouts supported. linear, fan-in, fan- out.<p>Future Features:<p>- Better examples for use with async calls (like LLMs)<p>- Splitting dataclasses, better error handling, logging improvements.<p>This is under active development. Any feedback, interest, or contributions are appreciated. Thanks!<p>github link: <a href="https://github.com/mcminis1/mr-graph">https://github.com/mcminis1/mr-graph</a>
Show HN: Ingest data from your customers (Prequel YC W21)
Hey HN! Charles here from Prequel (https://prequel.co). We just launched the ability for companies to import data from their customer’s data warehouse or database, and we wanted to share a little bit more about it with the community.<p>If you just want to see how it works, here’s a demo of the product that Conor recorded: https://www.loom.com/share/4724fb62583e41a9ba1a636fc8ea92f1.<p>Quick background on us: we help companies integrate with their customer’s data warehouse or database. We’ve been busy helping companies export data to their customers – we’re currently syncing over 40bn rows per month on behalf of companies. But folks kept on asking us if we could help them import data from their customers too. They wanted the ability to offer a 1st-party reverse ETL to their customers, similar to the 1st-party ETL capability we already helped them offer. So we built that product, and here we are.<p>Why would people want to import data? There are actually plenty of use-cases here. Imagine a usage-based billing company that needs to get a daily pull from its customers of all the billing events that happened, so that they can generate relevant invoices. Or a fraud detection company who needs to get the latest transaction data from its customers so it can appropriately mark fraudulent ones.<p>There’s no great way to import customer data currently. Typically, people solve this one of two ways today. One is they import data via CSV. This works well enough, but it requires ongoing work on the part of the customer: they need to put a CSV together, and upload it to the right place on a daily/weekly/monthly basis. This is painful and time-consuming, especially for data that needs to be continuously imported. Another one is companies make the customer write custom code to feed data to their API. This requires the customer to do a bunch of solutions engineering work just to get started using the product – which is a suboptimal onboarding experience.<p>So instead, we let the customer connect their database or data warehouse and we pull data directly from there, on an ongoing basis. They select which tables to import (and potentially map some columns to required fields), and that’s it. The setup only takes 5 minutes, and requires no ongoing work. We feel like that’s the kind of experience every company should provide when onboarding a new customer.<p>Importing all this data continuously is non-trivial, but thankfully we can actually reuse 95% of the infrastructure we built for data exports. It turns out our core transfer logic remains pretty much exactly the same, and all we had to do was ship new CRUD endpoints in our API layer to let users configure their source/destination. As a brief reminder about our stack, we run a GoLang backend and Typescript/React frontend on k8s.<p>In terms of technical design, the most challenging decisions we have to make are around making database’s type-systems play nicely with each other (kind of an evergreen problem really). For imports, we allow the data recipient to specify whether they want to receive this data as JSON blob, or as a nicely typed table. If they choose the latter, they specify exactly which columns they’re expecting, as well as what type guarantees those should uphold. We’re also working on the ability to feed that data directly into an API endpoint, and adding post-ingestion validation logic.<p>We’ve mentioned this before but it bears worth repeating. We know that security and privacy are paramount here. We're SOC 2 Type II certified, and we go through annual white-box pentests to make sure that all our code is up to snuff. We never store any of the data anywhere on our servers. Finally, we offer on-prem deployments, so data never even has to touch our servers if our customers don't want it to.<p>We’re really stoked to be sharing this with the community. We’ll be hanging out here for most of the day, but you can also reach us at hn (at) prequel.co if you have any questions!
Show HN: AI explanations for other people’s code
Hi HN,<p>As a developer, I much prefer to write code than read other people’s code. But most projects I work on involve other developers so it’s hard to avoid. Often I find it really hard to quickly parse other people’s code so I thought maybe ChatGPT could help.<p>ChatGPT does a really nice job of giving clear explanations when you paste in code and ask it to explain it. But the interface is a bit of a pain to use if you’re using it many times a day. It’s also hard to share explanations with co-workers. So I built whatdoesthiscodedo.com. Just paste your code and get your ChatGPT explanation and a sharable link you can give to coworkers.<p>I’m hopeful it can also be helpful to folks who aren’t professional developers. My co-founder at my main company, Beam Analytics, is more of a product guy so only needs to write scripts etc. Before, he’d often just find code online to copy and then struggle to make it work. But with this tool, he says he’s getting better intuition in understanding the code he’s using, and how to modify it.<p>We’d love your feedback. Email us at hi (at) beamanalytics.io or DM me on twitter @TheBuilderJR
Show HN: AI explanations for other people’s code
Hi HN,<p>As a developer, I much prefer to write code than read other people’s code. But most projects I work on involve other developers so it’s hard to avoid. Often I find it really hard to quickly parse other people’s code so I thought maybe ChatGPT could help.<p>ChatGPT does a really nice job of giving clear explanations when you paste in code and ask it to explain it. But the interface is a bit of a pain to use if you’re using it many times a day. It’s also hard to share explanations with co-workers. So I built whatdoesthiscodedo.com. Just paste your code and get your ChatGPT explanation and a sharable link you can give to coworkers.<p>I’m hopeful it can also be helpful to folks who aren’t professional developers. My co-founder at my main company, Beam Analytics, is more of a product guy so only needs to write scripts etc. Before, he’d often just find code online to copy and then struggle to make it work. But with this tool, he says he’s getting better intuition in understanding the code he’s using, and how to modify it.<p>We’d love your feedback. Email us at hi (at) beamanalytics.io or DM me on twitter @TheBuilderJR
Show HN: Modern Font Stacks – New system font stack CSS for modern OSs
Show HN: Modern Font Stacks – New system font stack CSS for modern OSs
Show HN: Django Developers – Reverse Job Board for Django Developers
Show HN: I made a self-hosted ChatGPT UI
Show HN: I made a self-hosted ChatGPT UI
Show HN: I made a self-hosted ChatGPT UI
Show HN: Using GPT-3 and Whisper to save doctors’ time
Hey HN,<p>We're Alex, Martin and Laurent. We previously founded Wit.ai (W14), which we sold to Facebook in 2015. Since 2019, we've been working on Nabla (<a href="https://nabla.com" rel="nofollow">https://nabla.com</a>), an intelligent assistant for health practitioners.<p>When GPT-3 was released in 2020, we investigated it's usage in a medical context[0], to mixed results.<p>Since then we’ve kept exploring opportunities at the intersection of healthcare and AI, and noticed that doctors spend am awful lot of time on medical documentation (writing clinical notes, updating their EHR, etc.).<p>Today, we're releasing Nabla Copilot, a Chrome extension generating clinical notes from video consultations, to address this problem.<p>You can try it out, without installation nor sign up, on our demo page: <a href="https://nabla.com/copilot-demo/" rel="nofollow">https://nabla.com/copilot-demo/</a><p>Here’s how it works under the hood:<p>- When a doctor starts a video consultation, our Chrome extension auto-starts itself and listens to the active tab as well as the doctor’s microphone.<p>- We then transcribe the consultation using a fine-tuned version of Whisper. We've trained Whisper with tens of thousands of hours of medical consultation and medical terms recordings, and we have now reached an error rate which is 3× lower than Google's Speech-To-Text.<p>- Once we have the transcript, we feed it to a heavily trained GPT-3, which generates a clinical note.<p>- We finally return the clinical note to the doctor through our Chrome extension, the doctor can copy it to their EHR, and send a version to the patient.<p>This allows doctors to be fully focused on their consultation, and saves them a lot time.<p>Next, we want to make this work for in-person consultation.<p>We also want to extract structured data (in the FHIR standard) from the clinical note, and feed it to the doctor’s EHR so that it is automatically added to the patient's record.<p>Happy to further discuss technical details in comments!<p>---<p>[0]: <a href="https://nabla.com/blog/gpt-3/" rel="nofollow">https://nabla.com/blog/gpt-3/</a>
Show HN: Using GPT-3 and Whisper to save doctors’ time
Hey HN,<p>We're Alex, Martin and Laurent. We previously founded Wit.ai (W14), which we sold to Facebook in 2015. Since 2019, we've been working on Nabla (<a href="https://nabla.com" rel="nofollow">https://nabla.com</a>), an intelligent assistant for health practitioners.<p>When GPT-3 was released in 2020, we investigated it's usage in a medical context[0], to mixed results.<p>Since then we’ve kept exploring opportunities at the intersection of healthcare and AI, and noticed that doctors spend am awful lot of time on medical documentation (writing clinical notes, updating their EHR, etc.).<p>Today, we're releasing Nabla Copilot, a Chrome extension generating clinical notes from video consultations, to address this problem.<p>You can try it out, without installation nor sign up, on our demo page: <a href="https://nabla.com/copilot-demo/" rel="nofollow">https://nabla.com/copilot-demo/</a><p>Here’s how it works under the hood:<p>- When a doctor starts a video consultation, our Chrome extension auto-starts itself and listens to the active tab as well as the doctor’s microphone.<p>- We then transcribe the consultation using a fine-tuned version of Whisper. We've trained Whisper with tens of thousands of hours of medical consultation and medical terms recordings, and we have now reached an error rate which is 3× lower than Google's Speech-To-Text.<p>- Once we have the transcript, we feed it to a heavily trained GPT-3, which generates a clinical note.<p>- We finally return the clinical note to the doctor through our Chrome extension, the doctor can copy it to their EHR, and send a version to the patient.<p>This allows doctors to be fully focused on their consultation, and saves them a lot time.<p>Next, we want to make this work for in-person consultation.<p>We also want to extract structured data (in the FHIR standard) from the clinical note, and feed it to the doctor’s EHR so that it is automatically added to the patient's record.<p>Happy to further discuss technical details in comments!<p>---<p>[0]: <a href="https://nabla.com/blog/gpt-3/" rel="nofollow">https://nabla.com/blog/gpt-3/</a>