I canceled my book deal
388 by azhenley | 244 comments on Hacker News.
ads
New best story on Hacker News: Show HN: 22 GB of Hacker News in SQLite
Show HN: 22 GB of Hacker News in SQLite
372 by keepamovin | 122 comments on Hacker News.
Community, All the HN belong to you. This is an archive of hacker news that fits in your browser. When I made HN Made of Primes I realized I could probably do this offline sqlite/wasm thing with the whole GBs of archive. The whole dataset. So I tried it, and this is it. Have Hacker News on your device. Go to this repo ( https://ift.tt/EYJNF8U ): you can download it. Big Query -> ETL -> npx serve docs - that's it. 20 years of HN arguments and beauty, can be yours forever. So they'll never die. Ever. It's the unkillable static archive of HN and it's your hands. That's my Year End gift to you all. Thank you for a wonderful year, have happy and wonderful 2026. make something of it.
372 by keepamovin | 122 comments on Hacker News.
Community, All the HN belong to you. This is an archive of hacker news that fits in your browser. When I made HN Made of Primes I realized I could probably do this offline sqlite/wasm thing with the whole GBs of archive. The whole dataset. So I tried it, and this is it. Have Hacker News on your device. Go to this repo ( https://ift.tt/EYJNF8U ): you can download it. Big Query -> ETL -> npx serve docs - that's it. 20 years of HN arguments and beauty, can be yours forever. So they'll never die. Ever. It's the unkillable static archive of HN and it's your hands. That's my Year End gift to you all. Thank you for a wonderful year, have happy and wonderful 2026. make something of it.
New best story on Hacker News: Show HN: Z80-μLM, a 'Conversational AI' That Fits in 40KB
Show HN: Z80-μLM, a 'Conversational AI' That Fits in 40KB
402 by quesomaster9000 | 92 comments on Hacker News.
How small can a language model be while still doing something useful? I wanted to find out, and had some spare time over the holidays. Z80-μLM is a character-level language model with 2-bit quantized weights ({-2,-1,0,+1}) that runs on a Z80 with 64KB RAM. The entire thing: inference, weights, chat UI, it all fits in a 40KB .COM file that you can run in a CP/M emulator and hopefully even real hardware! It won't write your emails, but it can be trained to play a stripped down version of 20 Questions, and is sometimes able to maintain the illusion of having simple but terse conversations with a distinct personality. -- The extreme constraints nerd-sniped me and forced interesting trade-offs: trigram hashing (typo-tolerant, loses word order), 16-bit integer math, and some careful massaging of the training data meant I could keep the examples 'interesting'. The key was quantization-aware training that accurately models the inference code limitations. The training loop runs both float and integer-quantized forward passes in parallel, scoring the model on how well its knowledge survives quantization. The weights are progressively pushed toward the 2-bit grid using straight-through estimators, with overflow penalties matching the Z80's 16-bit accumulator limits. By the end of training, the model has already adapted to its constraints, so no post-hoc quantization collapse. Eventually I ended up spending a few dollars on Claude API to generate 20 questions data (see examples/guess/GUESS.COM), I hope Anthropic won't send me a C&D for distilling their model against the ToS ;P But anyway, happy code-golf season everybody :)
402 by quesomaster9000 | 92 comments on Hacker News.
How small can a language model be while still doing something useful? I wanted to find out, and had some spare time over the holidays. Z80-μLM is a character-level language model with 2-bit quantized weights ({-2,-1,0,+1}) that runs on a Z80 with 64KB RAM. The entire thing: inference, weights, chat UI, it all fits in a 40KB .COM file that you can run in a CP/M emulator and hopefully even real hardware! It won't write your emails, but it can be trained to play a stripped down version of 20 Questions, and is sometimes able to maintain the illusion of having simple but terse conversations with a distinct personality. -- The extreme constraints nerd-sniped me and forced interesting trade-offs: trigram hashing (typo-tolerant, loses word order), 16-bit integer math, and some careful massaging of the training data meant I could keep the examples 'interesting'. The key was quantization-aware training that accurately models the inference code limitations. The training loop runs both float and integer-quantized forward passes in parallel, scoring the model on how well its knowledge survives quantization. The weights are progressively pushed toward the 2-bit grid using straight-through estimators, with overflow penalties matching the Z80's 16-bit accumulator limits. By the end of training, the model has already adapted to its constraints, so no post-hoc quantization collapse. Eventually I ended up spending a few dollars on Claude API to generate 20 questions data (see examples/guess/GUESS.COM), I hope Anthropic won't send me a C&D for distilling their model against the ToS ;P But anyway, happy code-golf season everybody :)
New best story on Hacker News: Exe.dev
Exe.dev
426 by achairapart | 280 comments on Hacker News.
https://ift.tt/bicH70o https://ift.tt/xiIaU8m https://ift.tt/saVZ2XP
426 by achairapart | 280 comments on Hacker News.
https://ift.tt/bicH70o https://ift.tt/xiIaU8m https://ift.tt/saVZ2XP
New best story on Hacker News: Show HN: Minimalist editor that lives in browser, stores everything in the URL
Show HN: Minimalist editor that lives in browser, stores everything in the URL
340 by medv | 111 comments on Hacker News.
I wanted to see how far I could go building a notes app using only what modern browsers already provide – no frameworks, no storage APIs, no build step. What it does: Single HTML file, no deps, 111 loc Notes live in the URL hash (shareable links!) Auto-compressed with CompressionStream Plain-text editor (contenteditable) History support Page title from first # heading Respects light/dark mode No storage, cookies, or tracking The entire app is the page source. https://textarea.my/
340 by medv | 111 comments on Hacker News.
I wanted to see how far I could go building a notes app using only what modern browsers already provide – no frameworks, no storage APIs, no build step. What it does: Single HTML file, no deps, 111 loc Notes live in the URL hash (shareable links!) Auto-compressed with CompressionStream Plain-text editor (contenteditable) History support Page title from first # heading Respects light/dark mode No storage, cookies, or tracking The entire app is the page source. https://textarea.my/
New best story on Hacker News: Nvidia buying AI chip startup Groq for about $20B in cash
Nvidia buying AI chip startup Groq for about $20B in cash
348 by nickrubin | 212 comments on Hacker News.
https://ift.tt/xwEDSHL...
348 by nickrubin | 212 comments on Hacker News.
https://ift.tt/xwEDSHL...
New best story on Hacker News: Tell HN: Merry Christmas
Tell HN: Merry Christmas
327 by basilikum | 107 comments on Hacker News.
Different cultures celebrate Christmas at different days and time zones are a thing. But it's Christmas here, so: Merry Christmas to everyone. I hope you get some rest and can spend time with people who are dear to you and get to focus on what's important rather than getting lost in stressing about everything having to be perfect. Also much love to everyone who cannot spend their Christmas with dear people. To make sure this post meets the relevancy criteria, here is a Wikipedia article about some Christmas (more precisely advent) tradition which I personally really like: https://ift.tt/zrov2lE
327 by basilikum | 107 comments on Hacker News.
Different cultures celebrate Christmas at different days and time zones are a thing. But it's Christmas here, so: Merry Christmas to everyone. I hope you get some rest and can spend time with people who are dear to you and get to focus on what's important rather than getting lost in stressing about everything having to be perfect. Also much love to everyone who cannot spend their Christmas with dear people. To make sure this post meets the relevancy criteria, here is a Wikipedia article about some Christmas (more precisely advent) tradition which I personally really like: https://ift.tt/zrov2lE
New best story on Hacker News: Some Epstein file redactions are being undone with hacks
Some Epstein file redactions are being undone with hacks
430 by vinni2 | 326 comments on Hacker News.
Related: https://ift.tt/cA6Dbhk https://ift.tt/DFimpce... https://ift.tt/WlgIazH...
430 by vinni2 | 326 comments on Hacker News.
Related: https://ift.tt/cA6Dbhk https://ift.tt/DFimpce... https://ift.tt/WlgIazH...
New best story on Hacker News: Flock Exposed Its AI-Powered Cameras to the Internet. We Tracked Ourselves
Flock Exposed Its AI-Powered Cameras to the Internet. We Tracked Ourselves
408 by chaps | 355 comments on Hacker News.
Archive Link: https://ift.tt/WxGUZ6h Also: https://www.youtube.com/watch?v=vU1-uiUlHTo – This Flock Camera Leak is like Netflix For Stalkers
408 by chaps | 355 comments on Hacker News.
Archive Link: https://ift.tt/WxGUZ6h Also: https://www.youtube.com/watch?v=vU1-uiUlHTo – This Flock Camera Leak is like Netflix For Stalkers
New best story on Hacker News: Show HN: Jmail – Google Suite for Epstein files
Show HN: Jmail – Google Suite for Epstein files
418 by lukeigel | 92 comments on Hacker News.
Hi everyone! My name's Luke and I made the original Jmail here alongside Riley Walz. We had a ton of friends collaborate on building out more of the app suite last night in lieue of DOJ's "Epstein files" release. Please AMA!
418 by lukeigel | 92 comments on Hacker News.
Hi everyone! My name's Luke and I made the original Jmail here alongside Riley Walz. We had a ton of friends collaborate on building out more of the app suite last night in lieue of DOJ's "Epstein files" release. Please AMA!
New best story on Hacker News: Computer animator and Amiga fanatic Dick van Dyke turns 100
Computer animator and Amiga fanatic Dick van Dyke turns 100
268 by ggm | 88 comments on Hacker News.
Here's a video from 2004 https://www.youtube.com/watch?v=Y1J9kfDCAmU It's his 100th birthday today.
268 by ggm | 88 comments on Hacker News.
Here's a video from 2004 https://www.youtube.com/watch?v=Y1J9kfDCAmU It's his 100th birthday today.
New best story on Hacker News: Ask HN: How can I get better at using AI for programming?
Ask HN: How can I get better at using AI for programming?
332 by lemonlime227 | 349 comments on Hacker News.
I've been working on a personal project recently, rewriting an old jQuery + Django project into SvelteKit. The main work is translating the UI templates into idiomatic SvelteKit while maintaining the original styling. This includes things like using semantic HTML instead of div-spamming, not wrapping divs in divs in divs, and replacing bootstrap with minimal tailwind. It also includes some logic refactors, to maintain the original functionality but rewritten to avoid years of code debt. Things like replacing templates using boolean flags for multiple views with composable Svelte components. I've had a fairly steady process for doing this: look at each route defined in Django, build out my `+page.server.ts`, and then split each major section of the page into a Svelte component with a matching Storybook story. It takes a lot of time to do this, since I have to ensure I'm not just copying the template but rather recreating it in a more idiomatic style. This kind of work seems like a great use case for AI assisted programming, but I've failed to use it effectively. At most, I can only get Claude Code to recreate some slightly less spaghetti code in Svelte. Simple prompting just isn't able to get AI's code quality within 90% of what I'd write by hand. Ideally, AI could get it's code to something I could review manually in 15-20 minutes, which would massively speed up the time spent on this project (right now it takes me 1-2 hours to properly translate a route). Do you guys have tips or suggestions on how to improve my efficiency and code quality with AI?
332 by lemonlime227 | 349 comments on Hacker News.
I've been working on a personal project recently, rewriting an old jQuery + Django project into SvelteKit. The main work is translating the UI templates into idiomatic SvelteKit while maintaining the original styling. This includes things like using semantic HTML instead of div-spamming, not wrapping divs in divs in divs, and replacing bootstrap with minimal tailwind. It also includes some logic refactors, to maintain the original functionality but rewritten to avoid years of code debt. Things like replacing templates using boolean flags for multiple views with composable Svelte components. I've had a fairly steady process for doing this: look at each route defined in Django, build out my `+page.server.ts`, and then split each major section of the page into a Svelte component with a matching Storybook story. It takes a lot of time to do this, since I have to ensure I'm not just copying the template but rather recreating it in a more idiomatic style. This kind of work seems like a great use case for AI assisted programming, but I've failed to use it effectively. At most, I can only get Claude Code to recreate some slightly less spaghetti code in Svelte. Simple prompting just isn't able to get AI's code quality within 90% of what I'd write by hand. Ideally, AI could get it's code to something I could review manually in 15-20 minutes, which would massively speed up the time spent on this project (right now it takes me 1-2 hours to properly translate a route). Do you guys have tips or suggestions on how to improve my efficiency and code quality with AI?
New best story on Hacker News: Australia begins enforcing world-first teen social media ban
Australia begins enforcing world-first teen social media ban
585 by chirau | 961 comments on Hacker News.
https://ift.tt/km1pJ96 https://ift.tt/iqpSyd8 https://ift.tt/otujwMX... ( https://ift.tt/wPQisFD )
585 by chirau | 961 comments on Hacker News.
https://ift.tt/km1pJ96 https://ift.tt/iqpSyd8 https://ift.tt/otujwMX... ( https://ift.tt/wPQisFD )
New best story on Hacker News: Valve reveals it’s the architect behind a push to bring Windows games to Arm
Valve reveals it’s the architect behind a push to bring Windows games to Arm
582 by evolve2k | 522 comments on Hacker News.
https://ift.tt/jMuEC0Q
582 by evolve2k | 522 comments on Hacker News.
https://ift.tt/jMuEC0Q
New best story on Hacker News: DeepSeek-v3.2: Pushing the frontier of open large language models [pdf]
DeepSeek-v3.2: Pushing the frontier of open large language models [pdf]
486 by pretext | 215 comments on Hacker News.
https://ift.tt/fZWlGDr https://ift.tt/A8cfJ35
486 by pretext | 215 comments on Hacker News.
https://ift.tt/fZWlGDr https://ift.tt/A8cfJ35
Subscribe to:
Comments (Atom)