News for hackers. Stuff that matters.

ryansworks

Re:Ask HN: What are you working on? (May 2025) on Sunday May 25, 2025 @07:06PM (#44092107)
Great point. I regret not being more systematic about this. I have tried on most popular models since gpt 3.5 launched, but it’s all been very ad hoc with the same general approach of building up a language feature by feature.
Re:Ask HN: What are you working on? (May 2025) on Sunday May 25, 2025 @05:47PM (#44091491)
It took about 10 hours. Arguably would have taken me 100 to do it manually but likely would have fewer bugs, there’s a few I’m aware of, but I bet there are many.

I started with a basic syntax for expressions and specified a lot up front such as it being a bytecode interpreter and using a recursive descent parser.

I found building it up feature by feature to be much more effective than one shotting an entire feature rich language. Still there was a lot of back and forth.

Re:Ask HN: What are you working on? (May 2025) on Sunday May 25, 2025 @05:37PM (#44091403)
Testing the limits of vibe coding. Created a programming language 100% via prompting a o4-mini-high, but did carefully review the code. https://github.com/ryanmcdermott/spress

Show HN: Spress – A vibe coded programming language

https://github.com/ryanmcdermott/spress

Way back when GPT 3.5 came out, I was looking for an interesting problem to test model capability. It needed to be one that is well represented in a model training set, but still requires some level of reasoning.After pondering it, I thought a simple dynamic programming language with a bytecode interpreter would fit the bill.The initial LLMs failed miserably, but every new generation I have...

( Read More... | no comments | on HN )
Re:Ask HN: Where would I find a specific image data set? on Thursday July 27, 2023 @12:50PM (#36896416)
Thank you I had not heard of that! I’ll check it out.

Ask HN: Where would I find a specific image data set?

I would like to train a model to identify a particular object class of a yellow jacket nest, specifically that of the Vespula squamosa species.I need a sufficient amount of images to train the network (probably 1000-2000). I've checked open source data sets like CIFAR and ImageNet, and there's nothing even close to this object class.What are some good resources for finding this very specific...

( Read More... | 2 comments | on HN )
I would tend to agree with you, though my tune might be different if I were an investment advisor!
Thank you! It was better than my second choice which was "Bird of Pair-a-Dice"
That's a very intriguing idea, I hadn't considered the possibilities of looking at public/private photos and sampling color values across to get some randomness. Private is certainly the key. With something like Birdseed, the randomness is totally public just as it is when getting randomness from atmospheric data. If someone figures out which wisps of clouds you are sampling or what search term you are using on Twitter, then the jig is up!
Re:Ask HN: What project are you working on? on Friday May 22, 2015 @12:37PM (#9589394)
I've been working on a project to learn Go. It's called Spotifind. It basically connects two musical artists by a chain of related artists using Spotify's API. Like the Kevin Bacon game, but for musicians. Here's a link to the project, still learning a lot!

https://github.com/ryanmcdermott/spotifind