Cahlen Humphreys PRO

cahlen

https://bigcompute.science

AI & ML interests

☠️💻

Recent Activity

repliedto victor's post about 3 hours ago

Want to share my enthusiasm for https://huggingface.co/zai-org/GLM-5.1 here too 🔥 I think we have it: our open source Claude Code = GLM-5.1 + Pi (https://pi.dev/) - Built a Three.js racing game to eval and it's extremely impressive. Thoughts: - One-shot car physics with real drift mechanics (this is hard) - My fav part: Awesome at self iterating (with no vision!) created 20+ Bun.WebView debugging tools to drive the car programmatically and read game state. Proved a winding bug with vector math without ever seeing the screen - 531-line racing AI in a single write: 4 personalities, curvature map, racing lines, tactical drifting. Built telemetry tools to compare player vs AI speed curves and data-tuned parameters - All assets from scratch: 3D models, procedural textures, sky shader, engine sounds, spatial AI audio! - Can do hard math: proved road normals pointed DOWN via vector cross products, computed track curvature normalized by arc length to tune AI cornering speed You are going to hear about this model a lot in the next months - open source let's go - and thanks z-ai🚀🚀

reacted to victor's post with 🔥 about 3 hours ago

liked a dataset about 4 hours ago

Roman1111111/claude-opus-4.6-10000x

View all activity

Organizations

Posts 4

Post

Huggingface just enabled cuda kernel repos!! This is crazy cool!

Expect a ton more portable number theory cuda kernels in the near future. I'm going to have a hell of a lot of fun with this new feature.

Appreciate it huggingface!

https://huggingface.co/kernels

Post

195

I was in the mood to fine tune a model, you know since we can just do shit these days, but I've also been on a pure math kick lately going back to some of my old academic stuff. And I remember specifically back then thinking about how difficult it was to narrow down a problem worth working on. This is hopefully help narrow that gap, at least for some niche parts of number theory.

Convergent is a model that knows number theory and computational mathematics. But I've also trained agentic tool calling loops, using prebuilt tools advertised on my computational number theory project website https://bigcompute.science

So essentially, you can use this to talk about unsolved problems, ways to attack them, how to write the cuda kernels to investigate them on your local GPU, or come up with brand new conjectures!

But most importantly it can zero you in on something that might be more interesting, especially if you have a model to talk with the problem about. I'm aware their aren't a ton of number theorists that just hangout together... heh.

So I'm sure there are bugs. Just post on github or huggingface and I'll be able to get to them if I have time.

Everything open source. Model weights, training data, training scripts. Have at it.

cahlen/Convergent-7B
cahlen/Convergent-7B-data
https://github.com/cahlen/convergent
https://bigcompute.science
https://mcp.bigcompute.science/mcp