
Hanchen Wang
@hcwww_
@Stanford & @Genentech, Jure & Aviv
ID: 1448803134
https://www.hanchenw.com 22-05-2013 12:32:03
291 Tweet
1,1K Followers
292 Following


Even the smartest LLMs can fail at basic multiturn communication Ask for grocery help → without asking where you live 🤦♀️ Ask to write articles → assumes your preferences 🤷🏻♀️ ⭐️CollabLLM (top 1%; oral ICML Conference) transforms LLMs from passive responders into active collaborators.



It's official - I am thrilled to share that I am joining MIT Biology, the Koch Institute at MIT, and MIT IMES as an Assistant Professor this fall! We will be a hybrid lab combining new technologies and computation to study ecDNA and tumor evolution. thejoneslaboratory.com

Sara Mostafavi (Genentech ) & I (Stanford University) r excited to announce co-advised postdoc positions for candidates with deep expertise in ML for bio (especially sequence to function models, causal perturbational models & single cell models). See details below. Pls RT 1/


The Arc Institute Virtual Cell Challenge: prizes worth up to $100,000 for accurately predicting cellular responses to genetic or chemical perturbations.







Join the next #gestalt seminar by Hanchen Wang this Friday- July 20 @10am EST globalspatial.org/seminar-series/ Dr Wang talks about SpatialAgent- #AI tool for #spatialbiology- using LLM & adaptive reasoning for expt design & #multimodal #spatialomic data analysis Luciano Martelotto 🛠🧬💻🇦🇺 Ioannis Vlachos // @ioavlachos.bsky.social

HLE has recently become the benchmark to beat for frontier agents. We FutureHouse took a closer look at the chem and bio questions and found about 30% of them are likely invalid based on our analysis and third-party PhD evaluations. 1/7


Want to join our efforts Microsoft Research AI for Science to push the frontier of AI for materials? We are the team behind MatterGen & MatterSim and we have 2 job openings! Each can be in Amsterdam, NL, Berlin, DE, or Cambridge, UK. It is a rare opportunity to join a highly talented,



After discussion with Jiaxin Shi, results from "Diffusion Beats Autoregressive in Data-Constrained Settings" look like an exploit of the AR model's overfitting. Without overfitting, there seems no hope for discrete diffusion to outperform AR; see the 10B token plot for example.
