The best Hacker News stories from Show from the past day
Latest posts:
Show HN: A search engine for your personal network of high-quality websites
Hey all,<p>Last time when we were on HackerNews [1], we received a lot of feedback, and we incorporated most of it.<p>- We have changed our name from grep.help to usegrasp.com<p>- A privacy policy page<p>- Bulk import<p>- Pricing page<p>We are happy to introduce a new feature: a personalized answer search engine that provides direct citations to the content on the page.<p>Demo: <a href="https://usegrasp.com/search?q=is+starship+fully+reusable?" rel="nofollow">https://usegrasp.com/search?q=is+starship+fully+reusable?</a><p>1 - <a href="https://news.ycombinator.com/item?id=35510949" rel="nofollow">https://news.ycombinator.com/item?id=35510949</a>
Show HN: A search engine for your personal network of high-quality websites
Hey all,<p>Last time when we were on HackerNews [1], we received a lot of feedback, and we incorporated most of it.<p>- We have changed our name from grep.help to usegrasp.com<p>- A privacy policy page<p>- Bulk import<p>- Pricing page<p>We are happy to introduce a new feature: a personalized answer search engine that provides direct citations to the content on the page.<p>Demo: <a href="https://usegrasp.com/search?q=is+starship+fully+reusable?" rel="nofollow">https://usegrasp.com/search?q=is+starship+fully+reusable?</a><p>1 - <a href="https://news.ycombinator.com/item?id=35510949" rel="nofollow">https://news.ycombinator.com/item?id=35510949</a>
Show HN: A search engine for your personal network of high-quality websites
Hey all,<p>Last time when we were on HackerNews [1], we received a lot of feedback, and we incorporated most of it.<p>- We have changed our name from grep.help to usegrasp.com<p>- A privacy policy page<p>- Bulk import<p>- Pricing page<p>We are happy to introduce a new feature: a personalized answer search engine that provides direct citations to the content on the page.<p>Demo: <a href="https://usegrasp.com/search?q=is+starship+fully+reusable?" rel="nofollow">https://usegrasp.com/search?q=is+starship+fully+reusable?</a><p>1 - <a href="https://news.ycombinator.com/item?id=35510949" rel="nofollow">https://news.ycombinator.com/item?id=35510949</a>
Show HN: Marqt.org lets you vote on the truth
I'm Arthur, and I wanted to share an MVP for Marqt.org, which lets you crowd-source the truth.<p>John Stuart Mill said that "Truth emerges from the clash of ideas."
In that spirit, Marqt brings two adversarial sides together to quantify truth and showcase the best arguments for each side.<p>It is inspired by markets, where buyers and sellers discover a product's true price and update it dynamically.
The ultimate aim is to build an open-source semantic knowledge base that represents the collective wisdom of humanity in real-time.<p>If we can do this, I believe it can solve the problem of misinformation in the media and slow the virality of disinformation.
More importantly, I believe can be a key part of the AGI stack and allow humans to be paid for building and maintaining it.<p>So what is a marqt?
A marqt is a short statement (120 chars or less) that can be true or false.
You swipe right to "marq it" true or swipe left to "marq it" false (or use hotkeys j/k or t/f).<p>Your "marq" counts toward a collective tally that determines how true or false any given marqt is (so something can be "68% true" or "93% false").
Unlike polls, marqts don't expire, so you can change your marq if new data or evidence comes to light or if you change your mind.<p>After you "make your marq," you are encouraged to add a "remarq," or an explanation for your vote.
Existing remarqs can be upvoted or downvoted based on how "remarqable" or "unremarqable" they are, a la Stack Overflow.<p>You can also "make a marqt" by entering a statement into the marqt maker. You submit by marqing it true or false.
It currently only accepts statements in the present tense ("current truths"), but "future truths" (predictions) will come down the line.
Inside the marqt maker is a fine-tuned BERT model that roughly ensures the statement is sensical, grammatical, and not a question.<p>"Show me the incentives and I will show you the outcome." -Charlie Munger<p>The incentives in betting markets are to win money, but for every winner there is a loser on the other side, making it arguably a zero-sum game.
The incentives in the marqt are to create engaging marqts that encourage the most participation through a point system called "marqs earned."
The goal is eventually for people to be able to trade in their marqs earned for money (via a "marqtplace," of course).<p>You earn one point for every marq you make, every remarq you add, every remarqable or unremarqable vote, and for every marqt you make.
However, if you are the marqt maker, you collect all the points from the activity in your marqt – every marq, remarq, and remarqable/unremarqable vote.
You also collect all the remarqable upvotes for your remarqs. Likewise, you lose points for unremarqable downvotes.<p>You can view everything and interact with the UI and inference model in demo mode (without signing up), but I hope you sign up for an account. I intentionally am not using analytics or cookies because I'm not interested in secretly tracking what users do on the site, I'm interested in showcasing what they create.<p>It's just an MVP for now. You may need to refresh the page every now and then, but please point out any bugs and feature requests.
I specifically learned how to program a year and a half ago so that I could build Marqt.org because I believe it is important for society to accurately define truth in an age when information travels (and changes) faster than ever, and when LLMs so confidently lie and have been heavily prompted to avoid controversial topics.<p>I have deep respect for the HN community, and I won't hide the fact that it's nerve-wracking to finally announce this because you're all so brilliant. I welcome your criticism and advice, and sincerely thank you for your time.<p>Arthur
arthur@marqt.org<p>Front-end: JavaScript
Back-end and API: Django/DRF
ML: deberta-v3-xsmall via Hugging Face
Deployed on: Google Cloud Run
Show HN: Marqt.org lets you vote on the truth
I'm Arthur, and I wanted to share an MVP for Marqt.org, which lets you crowd-source the truth.<p>John Stuart Mill said that "Truth emerges from the clash of ideas."
In that spirit, Marqt brings two adversarial sides together to quantify truth and showcase the best arguments for each side.<p>It is inspired by markets, where buyers and sellers discover a product's true price and update it dynamically.
The ultimate aim is to build an open-source semantic knowledge base that represents the collective wisdom of humanity in real-time.<p>If we can do this, I believe it can solve the problem of misinformation in the media and slow the virality of disinformation.
More importantly, I believe can be a key part of the AGI stack and allow humans to be paid for building and maintaining it.<p>So what is a marqt?
A marqt is a short statement (120 chars or less) that can be true or false.
You swipe right to "marq it" true or swipe left to "marq it" false (or use hotkeys j/k or t/f).<p>Your "marq" counts toward a collective tally that determines how true or false any given marqt is (so something can be "68% true" or "93% false").
Unlike polls, marqts don't expire, so you can change your marq if new data or evidence comes to light or if you change your mind.<p>After you "make your marq," you are encouraged to add a "remarq," or an explanation for your vote.
Existing remarqs can be upvoted or downvoted based on how "remarqable" or "unremarqable" they are, a la Stack Overflow.<p>You can also "make a marqt" by entering a statement into the marqt maker. You submit by marqing it true or false.
It currently only accepts statements in the present tense ("current truths"), but "future truths" (predictions) will come down the line.
Inside the marqt maker is a fine-tuned BERT model that roughly ensures the statement is sensical, grammatical, and not a question.<p>"Show me the incentives and I will show you the outcome." -Charlie Munger<p>The incentives in betting markets are to win money, but for every winner there is a loser on the other side, making it arguably a zero-sum game.
The incentives in the marqt are to create engaging marqts that encourage the most participation through a point system called "marqs earned."
The goal is eventually for people to be able to trade in their marqs earned for money (via a "marqtplace," of course).<p>You earn one point for every marq you make, every remarq you add, every remarqable or unremarqable vote, and for every marqt you make.
However, if you are the marqt maker, you collect all the points from the activity in your marqt – every marq, remarq, and remarqable/unremarqable vote.
You also collect all the remarqable upvotes for your remarqs. Likewise, you lose points for unremarqable downvotes.<p>You can view everything and interact with the UI and inference model in demo mode (without signing up), but I hope you sign up for an account. I intentionally am not using analytics or cookies because I'm not interested in secretly tracking what users do on the site, I'm interested in showcasing what they create.<p>It's just an MVP for now. You may need to refresh the page every now and then, but please point out any bugs and feature requests.
I specifically learned how to program a year and a half ago so that I could build Marqt.org because I believe it is important for society to accurately define truth in an age when information travels (and changes) faster than ever, and when LLMs so confidently lie and have been heavily prompted to avoid controversial topics.<p>I have deep respect for the HN community, and I won't hide the fact that it's nerve-wracking to finally announce this because you're all so brilliant. I welcome your criticism and advice, and sincerely thank you for your time.<p>Arthur
arthur@marqt.org<p>Front-end: JavaScript
Back-end and API: Django/DRF
ML: deberta-v3-xsmall via Hugging Face
Deployed on: Google Cloud Run
Show HN: Narr – Download Netflix audio for sampling
Show HN: Narr – Download Netflix audio for sampling
Show HN: Narr – Download Netflix audio for sampling
Shatter, the First Comic Made on a Computer (1985)
A comic book created by Mike Saenz, Peter B. Gillis and Charles Athanas in the mid-1980s on an Apple Macintosh (though with traditional, analog coloring). This milestone of the comics industry started out successfully but ran its course after 14 issues. It paved the way for other digital comics like Das Robot Imperium by Michael Götze, Iron Man: Crash by Mike Saenz and Batman: Digital Justice by Pepe Moreno.
Show HN: Ask Harry Potter any question with GPT-4
I've enjoyed using CharacterAI a lot, and I also use OpenAI's API's for work and personal projects. I wanted to see if I could get the model to behave as believably as CharacterAI counterparts with just a system prompt - and I think it does. Curious if others agree.
Show HN: GPT-JSON – Structured and typehinted GPT responses in Python
Hey HN, I've been using GPT a lot lately in some side projects around data generation and benchmarking. During the course of prompt tuning I ended up with a pretty complicated request: the value that I was looking for, an explanation, a criticism, etc. JSON was the most natural output format for this but results would often be broken, have wrong types, or contain missing fields.<p>There's been some positive movement in this space, like with jsonformer (<a href="https://github.com/1rgs/jsonformer">https://github.com/1rgs/jsonformer</a>) the other day. But nothing that was plug and play with GPT.<p>This library consolidates the separate logic that I built across 5 different projects. It lets you prompt the model for how it should return fields, inject variable prompts, handle common formatting errors, then cast to pydantic when you're done for typehinting and validation in your IDE. If you're able to play around with it, let me know what you think.
Show HN: GPT-JSON – Structured and typehinted GPT responses in Python
Hey HN, I've been using GPT a lot lately in some side projects around data generation and benchmarking. During the course of prompt tuning I ended up with a pretty complicated request: the value that I was looking for, an explanation, a criticism, etc. JSON was the most natural output format for this but results would often be broken, have wrong types, or contain missing fields.<p>There's been some positive movement in this space, like with jsonformer (<a href="https://github.com/1rgs/jsonformer">https://github.com/1rgs/jsonformer</a>) the other day. But nothing that was plug and play with GPT.<p>This library consolidates the separate logic that I built across 5 different projects. It lets you prompt the model for how it should return fields, inject variable prompts, handle common formatting errors, then cast to pydantic when you're done for typehinting and validation in your IDE. If you're able to play around with it, let me know what you think.
Show HN: GPT-JSON – Structured and typehinted GPT responses in Python
Hey HN, I've been using GPT a lot lately in some side projects around data generation and benchmarking. During the course of prompt tuning I ended up with a pretty complicated request: the value that I was looking for, an explanation, a criticism, etc. JSON was the most natural output format for this but results would often be broken, have wrong types, or contain missing fields.<p>There's been some positive movement in this space, like with jsonformer (<a href="https://github.com/1rgs/jsonformer">https://github.com/1rgs/jsonformer</a>) the other day. But nothing that was plug and play with GPT.<p>This library consolidates the separate logic that I built across 5 different projects. It lets you prompt the model for how it should return fields, inject variable prompts, handle common formatting errors, then cast to pydantic when you're done for typehinting and validation in your IDE. If you're able to play around with it, let me know what you think.
Show HN: GPT-JSON – Structured and typehinted GPT responses in Python
Hey HN, I've been using GPT a lot lately in some side projects around data generation and benchmarking. During the course of prompt tuning I ended up with a pretty complicated request: the value that I was looking for, an explanation, a criticism, etc. JSON was the most natural output format for this but results would often be broken, have wrong types, or contain missing fields.<p>There's been some positive movement in this space, like with jsonformer (<a href="https://github.com/1rgs/jsonformer">https://github.com/1rgs/jsonformer</a>) the other day. But nothing that was plug and play with GPT.<p>This library consolidates the separate logic that I built across 5 different projects. It lets you prompt the model for how it should return fields, inject variable prompts, handle common formatting errors, then cast to pydantic when you're done for typehinting and validation in your IDE. If you're able to play around with it, let me know what you think.
Show HN: Promptfoo – CLI for testing & improving LLM prompt quality
Show HN: Promptfoo – CLI for testing & improving LLM prompt quality
Show HN: USearch – Smaller and Faster Single-File Vector Search Engine
Last week was insane for vector search. Weaviate raised $50M, and Pinecone raised $100M... That's a lot and makes you believe that vector search is hard. But it's not.<p>I have spent the last couple of days implementing a single-file vector search engine from scratch, which is at least the tenth in the twenty years of my career. But this time, it's different. Instead of inventing a brand new algorithm and doing some crazy optimizations on the GPU, I:<p>1. took the standard HNSW algorithm,
2. fitted into 1000 lines of C++11 for portability,
3. added quantization and hardware-accelerated metrics,
4. wrapped for Python, JavaScript, Rust, and Java, and
5. open-sourced it!<p>It was fun, and to my surprise, it performed well, reaching 300K QPS on Amazon c7g instances. I never had to use third-party vector search products, but the first testers of USearch suggested 3x performance improvement over their existing solutions.<p>My colleagues and friends are also adding bindings for GoLang and the Wolfram language. We will soon add Windows support, a standalone server, and a distributed version based on UCall we shared a month ago. There are, of course, but you can already use it!<p>One of the apparent use cases is Semantic Search platforms. The example at the end of the GitHub page shows how to use USearch, UCall, and the UForm transformers together to build up a text-to-image semantic search platform in just 20 lines of Python.<p>Try it and join the development! We also have a lot of open positions, especially for those who want to work with us on next-get algorithms and AI infra rather than polishing and repackaging existing ideas :)
Show HN: USearch – Smaller and Faster Single-File Vector Search Engine
Last week was insane for vector search. Weaviate raised $50M, and Pinecone raised $100M... That's a lot and makes you believe that vector search is hard. But it's not.<p>I have spent the last couple of days implementing a single-file vector search engine from scratch, which is at least the tenth in the twenty years of my career. But this time, it's different. Instead of inventing a brand new algorithm and doing some crazy optimizations on the GPU, I:<p>1. took the standard HNSW algorithm,
2. fitted into 1000 lines of C++11 for portability,
3. added quantization and hardware-accelerated metrics,
4. wrapped for Python, JavaScript, Rust, and Java, and
5. open-sourced it!<p>It was fun, and to my surprise, it performed well, reaching 300K QPS on Amazon c7g instances. I never had to use third-party vector search products, but the first testers of USearch suggested 3x performance improvement over their existing solutions.<p>My colleagues and friends are also adding bindings for GoLang and the Wolfram language. We will soon add Windows support, a standalone server, and a distributed version based on UCall we shared a month ago. There are, of course, but you can already use it!<p>One of the apparent use cases is Semantic Search platforms. The example at the end of the GitHub page shows how to use USearch, UCall, and the UForm transformers together to build up a text-to-image semantic search platform in just 20 lines of Python.<p>Try it and join the development! We also have a lot of open positions, especially for those who want to work with us on next-get algorithms and AI infra rather than polishing and repackaging existing ideas :)
Show HN: Niui 3.0 – lightweight, rich, accessible front end
Here is a library of the most common components I've created in the last decade. It aims to solve the toughest UI problems like Carousel, Modal and Select, while using native browser capabilities as much as possible, and focusing on accessibility, stability and customisation. 14 KB of CSS, JS optional.<p><a href="https://rado.bg/niui-3-0-native-internet-user-interface/" rel="nofollow">https://rado.bg/niui-3-0-native-internet-user-interface/</a>
Show HN: Niui 3.0 – lightweight, rich, accessible front end
Here is a library of the most common components I've created in the last decade. It aims to solve the toughest UI problems like Carousel, Modal and Select, while using native browser capabilities as much as possible, and focusing on accessibility, stability and customisation. 14 KB of CSS, JS optional.<p><a href="https://rado.bg/niui-3-0-native-internet-user-interface/" rel="nofollow">https://rado.bg/niui-3-0-native-internet-user-interface/</a>