3 ms·
https://gist.github.com/cpsquonk/e9a6134e78a2c832161ca973803820e6 https://gist.github.com/cpsquonk/e9a6134e78a2c832161ca973803... I did Qwen3-256B (a free mode
by cap11235 1y ago
https://gist.github.com/cpsquonk/e9a6134e78a2c832161ca973803820e6 https://gist.github.com/cpsquonk/e9a6134e78a2c832161ca973803...
I did Qwen3-256B (a free model, but you'd need a host for something that large, probably. I used Kagi) and Claude Code.
Curious how these look to you.
- Animats 1y agoIt actually wrote out the code for all the hard stuff. I like the Python code which outsourced the hard stuff to existing libraries. The odds of that working are higher. Can you tell it to use the "glam" crate for the vectors, instead of writing out things like vector length the long way? (We now need standardized low-level types more than ever, so the LLMs will use them.)
- cap11235 1y agoI reopened Claude, and asked "Can you use the "glam" crate for the vectors, instead of writing out things like vector length the long way?" https://gist.github.com/cpsquonk/348009eb7c83a7d499ff5ae70d76c50e https://gist.github.com/cpsquonk/348009eb7c83a7d499ff5ae70d7...
- Animats 1y agoThat's pretty good. Thanks.