hammer32

4 ms·

1.
by hammer32 6mo ago
The training loop is heavy on floating-point math and string parsing, so it should exercise the JIT nicely. I'd love to hear how it performs!
2.
by hammer32 6mo ago
Thanks! The precision was a happy surprises, HyperTalk uses Apple's SANE library, which gives you 80-bit extended precision. The interpreter speed and the lack of arrays were a challenge. Rediscovering what HyperCard could do was hal
3.
by hammer32 6mo ago
Building this definitely felt like constructing a lightsaber from spare parts: slow, deliberate, but it works and you understand every piece of it.
4.
by hammer32 6mo ago
HyperTalk is an interpreted scripting language. The scripts are stored as plain text inside the stack and interpreted at runtime. It's kind of like a Visual Basic form where the UI and the code live in the same file. You can open an
5.
by hammer32 6mo ago
The stack is the code. You can view it directly for each button or examine the per-page script. As far as I know there isn't a compiler that lets you write standalone code and turn it into a stack. The stacks are dropped into Disk C
6.
by hammer32 6mo ago
Exactly. Working in a constrained environment invites innovation.
7.
by hammer32 6mo ago
Thank you! The constraints made it interesting. HyperCard doesn't have arrays, so the entire model, weights, activations, gradients, is stored as strings in hidden fields. All of the matrix math is done with "item i of field&qu
8.
by hammer32 6mo ago
Yup, that would have been easier. It's been decades since I've done anything with HyperCard. I had to re-take the built-in intro course again :)
9.
by hammer32 6mo ago
More of a copy-paste process. The scripts are written as .txt files in Nova on my Mac Studio, then pasted one at a time into HyperCard's script editor on the classic Mac. The files are kept separate because SimpleText has a 32 KB text
10.
by hammer32 6mo ago
Thanks! The quickest way to try it is the HyperCard Simulator link someone just posted in this thread: https://hcsimulator.com/imports/MacMind---Trained-69E0132C — go to the Inference card, click New Random to fill in
11.
by hammer32 6mo ago
I had no idea your simulator existed. No XCMDs, correct; everything is pure HyperTalk. I just ran a few training steps and they complete in a second or two. Thank you for importing it!
12.
by hammer32 6mo ago
Right? Backprop was published in 1986, a year before HyperCard shipped. Attention is newer, but a small model like this was buildable.
13.
Show HN: MacMind – A transformer neural network in HyperCard on a 1989 Macintosh(github.com)
159 points by hammer32 6mo ago|42 comments
14.
by hammer32 7mo ago
I built a system that lets AI agents (like Claude) interact with Classic Mac OS System 7 through OS 9 on real hardware and emulators. The architecture is simple: AgentBridge is a native Mac app that polls an inbox folder for text-file comma
15.
Show HN: AgentBridge – Let AI agents control Classic Mac OS thru a shared folder(github.com)
3 points by hammer32 7mo ago|2 comments
16.
Show HN: Agent Gate – Execution authority for AI agents, vault-backed rollback(github.com)
1 points by hammer32 8mo ago|1 comments
17.
by hammer32 8mo ago
AI agents (Claude Code, LangChain, CrewAI, MCP) all follow the same pattern: agent outputs a structured tool call, client code executes it. That gap between proposed and executed is a natural interception point, and almost nobody is build