Bernardino Romera-Paredes
@ber24
Followers
833
Following
907
Media
75
Statuses
1K
Previously core team member of AlphaFold2, AlphaTensor, and FunSearch at Google DeepMind. Now starting something new
London
Joined July 2009
Amazing team, amazing work!.
After 1.5 years of work, I'm so excited to announce AlphaEvolve – our new LLM + evolution agent!.Learn more in the blog post: White paper PDF: (1/2)
0
2
11
RT @AnjaSurina: Excited to share our latest work on EvoTune, a novel method integrating LLM-guided evolutionary search and reinforcement le….
0
31
0
This was a fun planning challenge, I wish there were more like this one. Thanks for organising!.
The Beluga™ Competition results are in! .🏆Congrats to teams led by @ber24 , Daniel Gnad & Jean Jodeau!.Thanks also to the hundreds who registered & explored the challenge. 👉Check the winning teams .#TrustworthyAI #ExplainableAI
0
0
1
RT @pushmeet: Our team @GoogleDeepMind has released AlphaTensor-Quantum, a new method that improves quantum circuits by reducing T gates ne….
0
9
0
RT @PetarV_93: A clear step towards achieving my dream: building AI that assists competitive programmers 🧑💻. “This is an exciting approach….
0
59
0
RT @xamat: We also need more awareness of existing tools by organizations and governments. Please take a look at Google's Floodcast tool (h….
deepmind.google
Our state-of-the-art model delivers 10-day weather predictions at unprecedented accuracy in under one minute
0
2
0
Great episode! Asked whether an evolutionary approach is needed to get to a higher level @_rockt said "Science, the way humans do it, is evolutionary search and I don't see any other way of how automated scientific process can work differently. It has to be evolutionary". 💯%.
I had a fantastic time talking with @samcharrington (@twimlai) about @orionbooks' "Artificial Intelligence: 10 Things You Should Know" ( and many exciting 2024 research papers (some of them from my teams) in the Open-Endedness community by outstanding.
2
0
13
Great to see FunSearch featured as one of this year’s top contributions to the field in the State of AI Report!.As always, a highly recommended report on how everything around AI keeps evolving. It even includes yesterday’s great news about the Nobel Prize awarded to Demis & John
🪩The @stateofaireport 2024 has landed! 🪩. Our seventh installment is our biggest and most comprehensive yet, covering everything you *need* to know about research, industry, safety and politics. As ever, here's my director’s cut (+ video tutorial!) 🧵
3
1
6
I'm excited to speak at the next ML in PL Conference in November, thanks @MLinPL for organising!.I'm planning to talk about our last work, FunSearch. If you have any questions or topics you’d like me to cover, related or not to FunSearch, let me know.
0
1
13
RT @ORonneberger: We just released AlphaFold 3! All details are in the paper and you can use it on our server!.
0
4
0
RT @JonasAAdler: While it's not "Ask LLM, get new knowledge", FunSearch relied heavily on LLMs to create new interesting and novel mathemat….
nature.com
Nature - FunSearch makes discoveries in established open problems using large language models by searching for programs describing how to solve a problem, rather than what the solution is.
0
1
0
Iba a escribir un hilo en español explicando FunSearch, el trabajo que hemos publicado hoy en Nature, pero @DotCSV ya lo ha hecho mejor de lo que yo podría:.
🔴 ¡IA programa AVANCES CIENTÍFICOS!. Hoy Google DeepMind ha presentado otro de esos trabajos que hoy son una chispa, y en un futuro algo muy brillante:. Una IA capaz de buscar e implementar funciones en código que sirvan de solución a problemas científicos!. Os cuento 🧵👇
8
15
115
I started FunSearch and can tell it wouldn't have gone far without this great team: @AlhusseinFawzi, @BarekatainAmin, @SashaVNovikov, @matejbalog, @Mpawankumar123, @emidup, @franciscuto, @JSEllenberg, @PengmingWang, Omar Fawzi, @pushmeet. The fun in FunSearch is really this team.
1
0
10
In this work we have striven for simplicity, creating a strong and simple method that we can build on in the future. This, together with LLMs getting more powerful, faster and cheaper make us highly confident about the groundbreaking potential FunSearch holds. 7/8.
1
1
3
2. When finding a tailored function for a practical use case, FunSearch solutions are easier to deploy than deep RL: easier to debug, more predictable and more scalable. 3. Importantly, users can get new insights about the problem and solution by looking at the code produced. 6/8.
1
2
4