Leonard Strouk (@_leostr_) header
Leonard Strouk (@_leostr_) profile picture

Co-founder and CTO of @livingmodels. Building in AI x Bio

Joined January 2022
If you want to learn more about how @jeandut14000, @VCabeli and the rest of the @livingmodels team are doing research :)
This one will feel different than my usual posts. Here I will not describe some fancy new ML model but instead I will describe both the research program and engineering systems we put in place at @livingmodels to make sure we can consistently deliver frontier models in biology=>
1
1
84
🧬 DNA 🧬 model factory babyyyyyy !
33
What actually happens inside a model lab between releases? @jeandut14000 , Alexey Rukhovich and @VCabeli wrote up how we approach big training runs at @livingmodels : deciding what’s worth a run, what we lock before launch, what we watch, what breaks. The part that never fits in a paper 👇 livingmodels.substack.com/p/…
1
1
1
6
316
If you’ve shipped one of these: what’s the thing you wish you’d known before your first big run?
1
27
Leonard Strouk retweeted
Open models for plant DNA analysis! 🍈 Combining Gemma 4 E4B + BOTANIC-1, a model trained on plant genomes, allowed the Living Models team to pinpoint which mutations on melon vines led to thriving, among many other cool experiments. Check it out!
12
14
1
124
14,040
Proud to see our work featured by @GoogleDeepMind 🌱 Google calls us an AI lab, but the result came from biologists and AI researchers working side by side, with guidance from advisors who run plant labs. That pairing is the whole point of @livingmodels . The setup: @googlegemma prepares and filters the genomic data, and BOTANIC-1, our genomic language model, scores the evolutionary impact of each candidate variant to find the causal mutation. In a recent test on melon, the pipeline ranked a known yield mutation #1, in an afternoon of compute. Finding variants like this used to take years of physical breeding. This is the direction we're building toward: models that shorten the path from genome to a climate-resilient crop. Thanks to the @GoogleDeepMind for the collaboration!
Finding the exact DNA that makes crops climate-resilient used to take years of physical breeding. To speed this up, AI lab Living Models pairs Gemma 4 E4B with BOTANIC-1, a genomic language model. Gemma prepares the genomic data, feeding filtered candidates to BOTANIC-1 to score evolutionary impact and pinpoint the causal mutation. In a recent test, they pinpointed a target melon yield mutation in an afternoon of computation, ranking it #1 out of 2,494 possibilities.
2
2
8
389
Really proud to communicate on what we are building at @livingmodels ! We paired @googlegemma with BOTANIC-1, our plant DNA genomic language model, and tested it on a task plant biologists, from private and public institutions, face in their daily lives !
2
1
4
226
Want to use it on your data, please reach out to [email protected] !
14
Leonard Strouk retweeted
🌱🌱🌱 a HUGE step towards designing to food of tomorrow -- from the DNA up 🧬🧬🧬 🙏 BOTANIC-1 just dropped, a family of models to do just that (beating all prev models) All open on @huggingface 🤗 + more useful than Millennium Problems More in 🧵
7
21
1
108
6,970
Millennium Prize Problems a making headlines, but your wheat 🌾 won't care ! Thrilled to introduce BOTANIC-1, our new plant DNA FM family. Outperform every model we tested on our benchmark. Weights: huggingface.co/collections/l… Paper: biorxiv.org/content/10.64898…
5
14
4
63
88,707
Better models should help researchers choose experiments that move the needle. We extended benchmarks to test variant prioritization from sequence, and used interpretability to probe gene structure - even flagging annotation errors! huggingface.co/spaces/living…
1
1
6
566
Three models with weights, available for non-commercial research. Test them. Build on them. Tell us where they break. @gsbenegas, @eczech0, @EdBuckler & fellow builders - we’d love your take 🍀 The story behind BOTANIC-1: livingmodels.substack.com/p/…
1
7
271
Leonard Strouk retweeted
Some personal news, a bit late. Sorry for the delay: I've been too busy cooking. Last October I left Mistral AI and joined Magic in SF to do frontier AI research. magic.dev/blog/pretraining
2
2
67
4,610
In Gainesville this week for the MOAI short course at @UF and proud that @livingmodels a sponsor. We'll present what we've been building in and pressure-test it with a sharp, diverse crowd. Come find me, I want the hard questions 😇 @Jarquin_GxE
1
2
75
Leonard Strouk retweeted
Everyone is building AI to generate text. We built AI to read DNA. Living Models is out of stealth. $7M raised. BOTANIC — our open-weight foundation model is live on Hugging Face today. The transformer behind ChatGPT doesn't care if it reads words or nucleotides. We do.
1
1
1
1
224