Daphne Cornelisse

(Multi-agent) simulation and RL • Ph.D. student NYU • Prev @Waymo @NVIDIAAI.

[Daphne Cornelisse @daphnesolves]

New Paper: Human-like Autonomy Emerges from Self-Play and a Pinch of Human Data.

We trained self-play RL on 60 years of simulation on 1 GPU in ~15 hours. Regularizing with 30 minutes of demonstration data produces much more human-like driving policies!

[Oded Rechavi @OdedRechavi]

When the boss motor protein tells another motor protein to come see him in his office

[Daphne Cornelisse @daphnesolves]

Sweden, here we come! 🇸🇪

I’m heading to ECCV, where we’ll host our workshop this Tuesday: https://t.co/GVo8UXckbo

We have a fantastic lineup of speakers; see you there!

[Michael Dennis @MichaelD1729]

world models offer new affordances for safety — like simulated counterfactuals to diagnose safety failures.

In this work, @Mingxuan0422 discovered we can use divide and conquer to scale counterfactual debugging to 1M steps

[Daphne Cornelisse @daphnesolves]

@momchiltomov Wow, this is really cool work! Congrats Momchil

[Momchil Tomov @momchiltomov]

Excited to share that our paper on explainable AI for autonomous driving is out today in @Nature !

Super proud to have been part of this amazing collaboration between @motionaldrive and @MIT_CSAIL!

Details in the 🧵

[Daphne Cornelisse @daphnesolves]

@learning_mech Wow, looks really cool! I like how the little humans are walking on the globe

[Daphne Cornelisse @daphnesolves]

Refactoring efforts pay off 🙂 (same functionality; less and more readable code)

[Daphne Cornelisse @daphnesolves]

An interesting read on the virtual cell.

[Sanjay Srivatsan @SRsrivatsan]

A new blog post thinking through the parallels between protein structure prediction and virtual cell efforts. Many of our current approaches to collect data to build a virtual cell model lack the abstraction that links measurement and function. (Link in reply)

[Daphne Cornelisse @daphnesolves]

@stevecrossan @tsuname Cool, sent you an email

[Daphne Cornelisse @daphnesolves]

btw, if you're working on molecular dynamics, drug design, or a related area and are curious about RL and high-performance simulation, please reach out! I'm looking for challenging problems in this space and would love to connect with people who bring domain expertise.

[Daphne Cornelisse @daphnesolves]

https://t.co/tAMkqIo4Xo

[Ryan Sullivan @RyanSullyvan]

Reward hacking has been in the news a lot lately, but AI researchers have seen surprising examples of it since long before LLMs. We're excited to share “AI Finds a Way,” led by @_aadharna , which brings many of these stories together in one place.

[Jim Beattie @JimBeattie18]

A good book, a warm cup
of coffee, and a little silence
- sometimes that's all the soul
needs to feel at home.

[Daphne Cornelisse @daphnesolves]

@EricSun50 What do you use to evaluate success here? Is it possible to do this in a simulation?

[Daphne Cornelisse @daphnesolves]

@tsuname @jchodera Interesting, and thanks for the recommendation! Will take a look at this.

[Daphne Cornelisse @daphnesolves]

@simonmaechling Sure, DMed you!

[Daphne Cornelisse @daphnesolves]

Absolutely. I would be naive to assume that this is a matter of finding the right objective and pressing go. I think several fundamental questions need to be answered first. For instance, 1) What knowledge do we have about system X to build the simulation around, and how reliable is it? 2) How do we formulate the problem? 3) How much knowledge about the system is "enough" for the simulation/policy to be useful (have predictive power) in the wet lab? I don't think this is an easy problem, but I find that an understanding of the practical challenges involved is a good foundation for doing science.

[Daphne Cornelisse @daphnesolves]

Building your dream home through trial and error. Really enjoyed this post - original thinking, as always: https://t.co/KvAyWacART

[Jakob Foerster @j_foerst]

TL;DR: This is one of the most important and exciting opportunities in AI on the planet - please read on.

The British Open-ended Learning & Discovery Lab is creating the perfect place for paradigm breaking AI research in the name of open-source and open-science. We have agency, we funding, we have unprecedented amounts of compute*, but WE NEED YOU!

..and we have created the dream job for you: The BOLD Fellow. This job combines a fast-moving, high agency, collaborative environment with full academic freedom and a salary that pays the bills.

Apply by noon UK time on the 15th of September for this once in a lifetime opportunity to shape the history of our field and of our planet:
https://t.co/wN009tQEkt

*by academic standards

[BioTender @Junioryu136689]

David Baker’s team is pushing AI protein design into its next phase. 🧬

AI models can now generate millions of protein designs — but the bottleneck has shifted from creation to validation.

In their latest Nature Communications paper, the Baker Lab develops a scalable experimental characterization platform to rapidly test AI-designed proteins and close the loop between Design → Build → Test → Learn.

The future of protein engineering may not just be bigger models — it may be faster feedback loops between AI and the wet lab.

Paper: https://t.co/aYcVOkGFck

[Daphne Cornelisse @daphnesolves]

@bern_jaeger Good news for KE:SAI!