Announcing AlphaEvolve, our new LLM coding agent that has - made new scientific discoveries - discovered algorithms that are now deployed at Google (in Gemini, Transformers, TPU hardware design & data centers) Blog: deepmind.google/discover/blo... White paper: storage.googleapis.com/deepmind-med... - ThreadSky

matejbalog.bsky.social • 33 days ago

Announcing AlphaEvolve, our new LLM coding agent that has
- made new scientific discoveries
- discovered algorithms that are now deployed at Google (in Gemini, Transformers, TPU hardware design & data centers)

Blog: https://deepmind.google/discover/blog/alphaevolve-a-gemini-powered-coding-agent-for-designing-advanced-algorithms/
White paper:
https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/alphaevolve-a-gemini-powered-coding-agent-for-designing-advanced-algorithms/AlphaEvolve.pdf

Comments

psyho.bsky.social•32 days ago

Did you try running it on some standard heuristic programming competitions? (i.e. atcoder's AHC or topcoder's marathons)

Would be nice to see it benchmarked vs something where the human effort can be easily quantified

ttkciar.bsky.social•33 days ago

Good for them. I proposed a similar thinger in 2023, but lacked the GPU-cycles to implement it (and didn't have the $$$ to buy GPU-cycles, either).

http://ciar.org/h/notes.cmc.txt

Maybe Gemma3-27B is good enough to implement CMC even without fine-tuning? I should find time to dork with it.

mehmetmars7.bsky.social•33 days ago

Surprised this is not shared as much here on bsky.

digthatdata.bsky.social•32 days ago

social sharing dynamics are not driven strictly by content value. people can be distracted by other content floating around the system, or there can be other reasons valuable information fails to go viral. This was just announced yesterday, give it a chance.

ttkciar.bsky.social•33 days ago

The anti-LLM contingent is strong on bsky. I'm more surprised that we're not blocked en masse for talking about it.

mathiastck.bsky.social•32 days ago

Expect some blocking, heck I expect commenting like this may get me a block or two.

mathiastck.bsky.social•32 days ago

For example, I have been previously added to a "Crypto/AI" moderation list:

https://bsky.app/profile/did:plc:i4jevytmqw4yg2vplnuyz6e3/lists/3kcykc7b3v62r

(I'm not at all a fan of crypto, and I am not interested in promoting "AI" just understanding its impact)

matejbalog.bsky.social•33 days ago

AlphaEvolve advanced over 10 open problems across Maths. But one is particularly precious to me:

It found the first improvement after 56 years to how fast we can multiply 4x4 matrices 🤯 (with complex number entries)

And it did so by discovering a sophisticated search algorithm

matejbalog.bsky.social•33 days ago

We’ve only scratched the surface of possible use cases of AlphaEvolve

I’m looking forward to many more applications across the sciences and computational problems

matejbalog.bsky.social•33 days ago

AlphaEvolve was developed by our wonderful team working on AI for algorithm discovery.

On BlueSky: @emidup.bsky.social @swarat.bsky.social

jakeforsey.bsky.social•26 days ago

This release inspired me! I went and read FunSearch, what an awesome and simple idea.

I implemented something inspired by it to optimise agents within my jax "swarm" environment. It works with only 3x1080 Ti running Qwen3-14B, amazing!
Repo https://github.com/JakeForsey/swarm
Best https://github.com/JakeForsey/swarm/blob/main/swarm/agents/vibevolve_v5.py

Comments

Posting Rules

Reply