←
Aurel Prosz279 posts
Aurel Prosz avatar

Aurel Prosz

@aurel_pr

Postdoc at Danish Cancer Institute. Interested in cancer genomics and applying LLMs in science. Follow me for news and repos I find online about AI and science.

◷ Joined December 2016▧ 35 Photos and videos
Aurel Prosz avatar
Aurel Prosz@aurel_pr·Jul 23

Yesterday I finally set up one of the largest, open-source LLM models in the world, GLM-5.2, on Gefion, the Danish supercomputer, to find new synthetic lethal relationships in cancer. It's going strong at around 5 tokens/second (I have to optimize the inference a bit...), evaluating a new gene pair every ~15 mins. Qualitatively, the reasoning capability of the model already looks better than that of Qwen 32B-instruct, but requires 2 full nodes, 2x8 H100 GPU-s to run it, and it's not quite trivial to set up the inter-node sharding. See comments for our manuscript, now in peer review.

0:10
Rohan Arun avatar
⟳ Aurel Prosz retweeted
Rohan Arun@RohanArun·Nov 1

Interesting idea: if we can formalize the body and real world into code with a graph representation, can agents solve problems with code? It seems to work.. I've been working on solving "impossible" problems with agents, and one possible path is to code a hyper accurate simulation of the human body to test medicine and proteins. However running it in real time would be very hard at the scale of atoms, so it's possible to use graphs and code to optimize it with agents just like we're doing with the auto-research Github repos. Open source repo to explore the idea here and agents welcome :) github.com/rohanarun/virtual…

Post attachmentPost attachmentPost attachmentPost attachment
Aurel Prosz avatar
Aurel Prosz@aurel_pr·Oct 27

Also here: I'm looking for any methods to generate negative energy and safeguards to not break reality itself by messing up causality.

Aurel Prosz avatar
Aurel Prosz@aurel_pr·Oct 27

Let's build a warp drive! Motivated by recent community efforts to use AI agents to tackle mathematical problems, I was wondering how to use the same framework to find solutions to grand, practical problems in physics. Point your agents at my repo and help us find more efficient warp bubbles travelling at 2C. My current one still requires 1400 Earth and anti-Earths to annihilate. The framework automatically checks physical constraints and validity.

0:20
Aurel Prosz avatar
Aurel Prosz@aurel_pr·Oct 26

The repo includes a proposed O*(2.33^k) algorithm, the proof, code, and research notes. Interested in checking the proof, finding counterexamples, or improving the algorithm? github.com/Paureel/rooted-fo…

Aurel Prosz avatar
Aurel Prosz@aurel_pr·Oct 26

Inspired by recent community math efforts, I've open-sourced my work on Maximum Agreement Forest: splitting evolutionary trees into the fewest matching pieces, useful for studying gene transfer between bacteria and hybridization in plants, among other applications. Point your agents towards my repo and help improve the current state of the art.

Post attachment
Aurel Prosz avatar
Aurel Prosz@aurel_pr·Oct 10

I'm experimenting with mapping out all the ongoing community efforts to improve, validate, or extend the mathematical results OpenAI released earlier this week. Check it out below! math-frontier-atlas.netlify.…

0:10
Aurel Prosz avatar
Aurel Prosz@aurel_pr·Oct 9

Thanks everyone for the feedback, the integer multiplication tracker now has a zoom feature, so hopefully you will be able to see the most recent progress better! Let me know in the comments if you have other ideas to add! beyond-n-log-n.netlify.app

Post attachment
Aurel Prosz avatar
Aurel Prosz@aurel_pr·Oct 8

I've just made a live tracker for the community effort for optimizing the "Integer multiplication below n log n" mathematical problem, so everybody can see the progress on it: beyond-n-log-n.netlify.app

Post attachment
Doug Colkitt avatar
⟳ Aurel Prosz retweeted
Doug Colkitt@0xdoug·Oct 8

Went to bed last night, and sorry to report only a bit of incremental progress on my side. *But* it appears like this has taken off in terms of now having a real community effort. Really inspiring to see what looks like major progress from some of you. I’m going to try to go through all the new contributions, validate the results, definitely make sure we have attribution, and coordinate a united front. I think I saw most of the contributions, but my notifications are a bit of a mess right now. If you haven’t, cutting a PR in the repo and/or just replying to this thread would help make sure we can count your work. For everyone who’s now tackling this problem as citizen scientists, awesome work guys. Really inspiring stuff. Let’s see how far we can push this

We are publishing an update to OpenAI problem #109 (integer multiplication) with further tightening. κ = 2⁻³⁴ (tightened from κ = 2⁻¹⁸²) The exact witness is 8.3 × 10⁻¹¹, a roughly 48 million fold improvement over the previous result and a 2¹⁴⁸ fold improvement over original OAI result. The improvement came from removing the spacing penalty behind the quadratic bottleneck. This was done by moving compact control bits instead of entire windows.