Hacker News: nohat

New comment by nohat in "Show HN: Spelling Riddle – Think Spellbee with crossword clue and visual hint"

nohat — Fri, 13 Feb 2026 15:33:00 +0000

Yep, I need to compare the new models and update everything. Foolish of me to generate ahead at all!

New comment by nohat in "Show HN: Spelling Riddle – Think Spellbee with crossword clue and visual hint"

nohat — Thu, 12 Feb 2026 23:59:49 +0000

Fair points, thank you for the very thoughtful feedback! "Alexandrian who found area from mere sides" is a bad hint because it is fairly obscure, and really just a history trivia check. And the image is a bit random. You can view the explanation for the hint by clicking the ? button in the upper right corner after a word is completed if you are curious about the models reasoning.

As a bit of an explanation I generated these before Nano banana pro came out, and at the time I made a large comparison grid for various image and text models. For this style qwen image performed very well. LLM wise I started with 5.1 and updated to 5.2. Of course with the rate of model release my choices are pretty much already obsolete... Expense is also a factor for a hobby project, and NB pro is 7.5x more expensive than Qwen image.

Show HN: Spelling Riddle – Think Spellbee with crossword clue and visual hint

nohat — Thu, 12 Feb 2026 19:00:22 +0000

I wanted a daily spelling game that tested semantic knowledge rather than vocab, and a riddle game with both text and visual hints. So I combined them.

Each day you get 9 letters and up to 15 hidden words to find. Every word has two orthogonal clues: a text hint and a image hint. The hints are crossword puzzle style — you need to figure out what the clues are pointing at, then spell it.

The site is fully static.

## How the clues get made

The pipeline has three stages:

1. Dictionary curation — Start from raw word lists, normalize, dedupe, filter plurals/profanity, and add frequency data.

2. Word quality tournament — Words are ranked for puzzle suitability using grouped LLM matches in a Swiss-style tournament. Performance is converted to a 0–100 quality rating using Wilson lower-bound scoring.

3. Hint generation — For each puzzle word set, generate multiple text + image hint candidates, lint-filter weak ones (e.g. text hints that contain the answer, image prompts that would render as text), rank the survivors, then generate the final images.

Text hints: GPT 5.2, Image hints: Qwen Image.

The clue generation is the hardest part and still not perfect (though model improvements might make it trivial soon).

Comments URL: https://news.ycombinator.com/item?id=46993364

Points: 2

# Comments: 4

New comment by nohat in "Teaching LLMs how to solid model"

nohat — Thu, 24 Apr 2025 04:42:23 +0000

Just a python script to render then api call with a prompt to check if the render looks right.

New comment by nohat in "Teaching LLMs how to solid model"

nohat — Wed, 23 Apr 2025 19:01:14 +0000

I tried this a few months back with claude 3.5 writing cadquery code in cline, with render photos for feedback. I got it to model a few simple things like terraforming mars city fairly nicely. However it still involved a fair bit of coaching. I wrote a simple script to automate the process more but it went off the rails too often.

I wonder if the models improved image understanding also lead to better spatial understanding.

New comment by nohat in "Igneous Linearizer: semi-structured source code"

nohat — Tue, 25 Jun 2024 22:09:27 +0000

Yeah, I definitely see using this for literate programming. Not quite sure the best way to organize it. Maybe use a static site compiler to auto host documentation version.

New comment by nohat in "'Stupid,' 'shameful:' Tech workers on Y Combinator CEO Garry Tan's rant"

nohat — Fri, 02 Feb 2024 17:41:08 +0000

It was a stupid and shameful tweet. The thing that elevates this above a drunk tweet though are the "real life" postcards. Should we ignore the fact that the postcards seem very much a false flag?

New comment by nohat in "Mistral CEO confirms 'leak' of new open source AI model nearing GPT4 performance"

nohat — Wed, 31 Jan 2024 20:04:43 +0000

It has been interesting seeing the sleuthing on this one. IMHO it is unfortunate to have this happen to a company that has been very pro open source.

New comment by nohat in "OpenAI board in discussions with Sam Altman to return as CEO"

nohat — Sun, 19 Nov 2023 01:58:35 +0000

The link just changed. Why? The original was the verge article, that was frankly terrible. It really read like the author had a specific goal.

New comment by nohat in "Show HN: Open Lecture, a ChatGPT Plugin that sources MIT lectures with timestamp"

nohat — Mon, 22 May 2023 16:45:55 +0000

Have you tried using ChatGPT to study? It's pretty incredible for learning, particularly interactive exploration. The point of this plugin is to give ChatGPT access to reliable information to help it be more accurate and informative, as well let it provide real sources so you can verify and continue learning. MIT OpenCourseWare is a great resource (the quality of the teachers is amazing), but it isn't easy to find what you want, and it lacks the interactive component of a real class. I made this to try combining the strengths of both.

Show HN: Open Lecture, a ChatGPT Plugin that sources MIT lectures with timestamp

nohat — Mon, 22 May 2023 16:26:51 +0000

I made Open Lecture, a plugin that gives ChatGPT access to MIT lectures and course materials. It helps Chatgpt give accurate responses and real, timestamped sources.

Demo: https://www.youtube.com/watch?v=xDJUgDprndk

Installation: In ChatGPT select GPT-4 -- select "Plugins" in the dropdown -- Click "Plugin Store" -- look for "Open Lecture".

Try it out, and let me know if you have any feedback, Thanks!

Comments URL: https://news.ycombinator.com/item?id=36033492

Points: 2

# Comments: 2

New comment by nohat in "Microsoft reportedly orders chatbot rivals to stop using Bing’s search data"

nohat — Sat, 25 Mar 2023 22:30:59 +0000

This is a shame as bing search api was much better documented, and supported (at least as of a few years ago). Google switched api specs and dropped features at an even faster rate than their consumer facing products.

New comment by nohat in "DeepMind has open-sourced the heart of AlphaGo and AlphaZero"

nohat — Wed, 15 Feb 2023 13:59:11 +0000

I feel like this headline is deceptive. This is a Jax reimplementation, and it was released a year ago. It is a cool library though. The basic operation of muzero is very simple, but training it efficiently is tricky.

New comment by nohat in "How Organisms Come to Know the World: Fundamental Limits on AGI"

nohat — Sun, 08 Jan 2023 15:41:22 +0000

We can define intelligence in a very real, practical way now. We see and identify intelligence all the time in humans and in animals and in AI. We may not be perfect at identifying it (just like grog might mistake a rising sun for a forest fire), but we don't need a perfect mathematical or philosophical definition that we all agree on to create it. We just need to rub sticks together really hard.

New comment by nohat in "How Organisms Come to Know the World: Fundamental Limits on AGI"

nohat — Sun, 08 Jan 2023 13:24:43 +0000

Mog: “What truly is fire? The divine blessing stolen by Prometheus? Concentrated Phlogiston? The element of change? Is it not madness to seek to create something we don’t even have a good definition of?”

Grog: “Grog rubs two sticks together” Lowers voice and looks around furtively “really hard.”

New comment by nohat in "Stability.ai sent a take down request to Runway ML's SD v1.5 citing IP Leak"

nohat — Thu, 20 Oct 2022 21:52:09 +0000

Maybe that post has been edited into coherence, but when it was posted that summary was misleading or flat wrong about a lot of points.

New comment by nohat in "Show HN: InvokeAI, an open source Stable Diffusion toolkit and WebUI"

nohat — Mon, 10 Oct 2022 19:19:15 +0000

I've been using a modified version of lsteins fork since almost the beginning. Recommended! It does lack some of the features of eg automatic1111, but it has good cli, and actually has a license, which is pretty important (as novelai has learned).

New comment by nohat in "Did GoogleAI just snooker one of Silicon Valley’s sharpest minds?"

nohat — Fri, 16 Sep 2022 12:04:32 +0000

I agree that declaring a win is a bit impolite _if_ the other person hasn't agreed. But changing "farmer" to "robot farmer" because Google won't allow him to generate pictures with humans is obviously not changing the goalposts in the usual meaning of the term.

New comment by nohat in "Correctness and composability bugs in the Julia ecosystem"

nohat — Mon, 16 May 2022 14:24:17 +0000

The power of allowing everyone to make foundational types and functions that work together is indeed dangerous. I'm not sure you are better off in the even more dangerous waters of c/c++/fortran, except that they are older and more established with many times the man-hours sunk into them. Is there a good way to control the interaction of these many different libraries with losing the generality and composability of Julia?

I will say that as a matter of language design 1 based indexing is perfectly fine, 0 based indexing is perfectly fine. Choose your own indexing is a hilarious foot gun, so no surprise it went off sometimes. Fortunately using it seems to be quite rare.

New comment by nohat in "Google doesn't want to pay Sonos for technology it copied"

nohat — Tue, 11 Jan 2022 16:57:14 +0000

How much did Sonos pay to get this article written? It's tripping over itself praising Sonos and repeating over and over that Google should pay Sonos.