Michelle Pan

sometimes i forget that people are listening… come tuesdays at 6pm 🙃

We asked Stanford students their best ML pickup lines 💌 for Valentine's Day 💝: Check out the full dataset 🔍: ml-valentines.github.io (wonderfully compiled by my friend @michelllepan) Here are some of the best ones (along with some doodles I made to match!)

(1/n) Check out our new paper Communicate to Play! 📜 We study cross-cultural communication and pragmatic reasoning in interactive gameplay 😃 Paper: arxiv.org/abs/2408.04900 Code: github.com/icwhite/codena… Talk: youtube.com/watch?v=r_tNbZ…

@mariah17schrum and @michelllepan presenting our work on RL for deep brain stimulation at ICML <3

Come visit @mariah17schrum and @michelllepan at the poster hall (#308) to learn more about this work at #ICML2024 !

something special about being able to touch your code

html energy was real (shoutout to the best cohost @upcycledwords)

Introducing our work on sample-efficient learning for adaptive brain stimulation! We use model-based RL to learn individualized neural coprocessor policies with minimal patient interactions 🧠 Grateful to have had amazing mentorship throughout this project—check it out below:

once again thinking about how the overleaf and whole foods logos are nearly mirror images of each other

did it hurt when you went through reverse diffusion? because you look like the SDXL output for “my soulmate”

the cloudy days in Berkeley are forgivable when the sunset looks like this :)

Preference-based reward learning (aka RLHF) is everywhere (robotics, recommenders, ChatGPT!). But does it always work? Our #ICLR2023 paper analyzes how PbRL can fail due to causal confusion and reward misidentification. Paper+Code: sites.google.com/view/causal-re… [1/n]

if your loss graph does something weird it's called a plot twist

Try out the Sieve platform to see just how much is possible, @mokshith_v and @thpicy are building something amazing :)