Theo Neukomm Profile
Theo Neukomm

@TheoNeukomm

Followers
78
Following
159
Media
4
Statuses
18

PhD candidate in LIAC at EPFL (@SchwallerGroup) - Machine Teacher I guess, driven by ML in chemistry and other domains of AI.

Lausanne, Switzerland
Joined May 2023
Don't wanna be here? Send us removal request.
@drecmb
Andrés M Bran
8 months
I'll be presenting at the #AI4Mat and #llmreasoning workshops, our work on reasoning in LLMs, with @TheoNeukomm 🧵: https://t.co/X8ecpyvNmD 📎: https://t.co/O894Im0sE6 🧑‍💻:
Tweet card summary image
github.com
Contribute to schwallergroup/steer development by creating an account on GitHub.
@drecmb
Andrés M Bran
9 months
LLMs are pretty bad at writing molecules, but quite good at analyzing mols and reactions! In our new work we use LLMs+search in chemical tasks, unlocking steerable synth. planning and mechanism prediction 🌟 1/ @TheoNeukomm @d_armstr @ZJoncev @pschwllr https://t.co/O894IlZUOy
1
2
3
@SchwallerGroup
LIAC at EPFL
8 months
Quick maths 🧮: In a lab of 25 people from 15+ countries, what's the probability that two share the same birthday —🗓️ 25 April— and they're both at #ICLR2025 🇸🇬? Drop your estimate below, or just wish @6ojaHa 🇷🇸 and @TheoNeukomm 🇨🇭 a huge happy birthday! 🎉🎂🥳🎁🍰
1
2
19
@efssh
Edvin Fako
8 months
🔥 AutoAdsorbate 🔥 has been the workhorse of high throughput computation in heterogeneous catalysis modeling and ML @BASF for the previous few years. 🧵 Now it is available open source! #MatSci #CompChem Preprint: https://t.co/KzkAzkKx0D Code: https://t.co/p0nfGeDNzc
1
20
59
@6ojaHa
BOjana Ranković
8 months
🚨New preprint! ✨My precious✨ GOLLuM: GP-Optimized LLMs for Bayesian Optimization — the first of its kind! LLMs as deep kernels in GPs jointly optimized via marginal likelihood → implicit metric learning, calibrated uncertainty & superior sampling 🚀 📄
Tweet card summary image
arxiv.org
Scientific discovery increasingly depends on efficient experimental optimization to navigate vast design spaces under time and resource constraints. Traditional approaches often require extensive...
4
18
79
@ZJoncev
Zlatko Jončev
8 months
Exciting new preprint! 🧪🤖 Our work shows how LLMs can steer search processes in chemistry! Great collaboration with: @drecmb @TheoNeukomm @d_armstr and @pschwllr in amazing @SchwallerGroup at @EPFL_en Check the demo for steerable retrosynthesis 🍿👇
1
7
22
@TheoNeukomm
Theo Neukomm
9 months
Finally, I’d like to thank my PI and coworkers, what an exciting time to be in the @SchwallerGroup! Special thanks as well to my sponsors @intel x @merckgroup! You want to know even more about our work? https://t.co/aRugjK34Zz Last/🧵
Tweet card summary image
arxiv.org
While automated chemical tools excel at specific tasks, they have struggled to capture the strategic thinking that characterizes expert chemical reasoning. Here we demonstrate that large language...
0
0
4
@TheoNeukomm
Theo Neukomm
9 months
Recently released LLMs were surprisingly good at adapting and understanding the task, even given this unusual environment to work with. 5/🧵
1
0
1
@TheoNeukomm
Theo Neukomm
9 months
The approach on mechanisms in this paper is to consider arrow pushing as a game that follows some rules, defining a state/action space and allowing the LLM to play the role of a critic, guiding the search with its chemical knowledge. 4/🧵
1
0
2
@TheoNeukomm
Theo Neukomm
9 months
In this work, we are showcasing how LLMs can be useful on multiple scales, by making them work on retrosynthesis ⬅️ , and on chemical mechanisms ⚙️. I am particularly happy to have worked on the mechanism part, where a lot of exciting research is being done. 3/🧵
1
0
2
@TheoNeukomm
Theo Neukomm
9 months
Have you ever struggled to encode a chemical concept into a computer even though you could simply explain it in words? Our Steer framework might be the exact tool you are looking for. Steer paves the way to a powerful use of LLMs with traditional search methods in chemistry. 2/🧵
1
0
1
@TheoNeukomm
Theo Neukomm
9 months
We just released a preprint with my wonderful coworkers @drecmb @d_armstr @ZJoncev @pschwllr showcasing how LLMs can refine a search in the chemical space using natural language as their compass. Preprint can be found at https://t.co/aRugjK34Zz, more on that below: 1/🧵
@drecmb
Andrés M Bran
9 months
LLMs are pretty bad at writing molecules, but quite good at analyzing mols and reactions! In our new work we use LLMs+search in chemical tasks, unlocking steerable synth. planning and mechanism prediction 🌟 1/ @TheoNeukomm @d_armstr @ZJoncev @pschwllr https://t.co/O894IlZUOy
2
4
16
@SchwallerGroup
LIAC at EPFL
1 year
LIAC has a newcomer, oh I mean Neukomm(er) Theo! @TheoNeukomm just defended his Master's thesis and is officially joining our lab in September as a PhD student! 🤩🤩 Thank you Theo for bringing your awesomeness to the lab alongside the free drinks 🧉🍻🥳
4
2
34
@gabepgomes
Gabe Gomes
2 years
the @SchwallerGroup 🤝 @gpggrp alliance commences with the arrival of @TheoNeukomm! bringing a bit of @EPFL_en to @CarnegieMellon.
0
4
45
@SchwallerGroup
LIAC at EPFL
2 years
🚀 The LIAC team is at #NeurIPS2023! Engage with our team's innovative work and check out our line-up for today! 👇
1
8
27
@6ojaHa
BOjana Ranković
2 years
Welcome to CHAOS! 🤯 Chemical additive optimization screaming 😱 using Bayesian optimization! Nо, not the Openai drama🍿but now that it's finally over (I hope) I can share our paper that just got published in Digital Discovery! 📜 https://t.co/s1j18ma0Cl 💻 https://t.co/wxD3J9VVyC
3
20
75