Besteuler's profile picture. Assistant Professor @CUHKofficial. Postdoc @MPI_IS. PhD @Cambridge_Uni & @GeorgiaTech. Previous Intern @Google & @nvidia. All opinions are my own.

Weiyang Liu

@Besteuler

Assistant Professor @CUHKofficial. Postdoc @MPI_IS. PhD @Cambridge_Uni & @GeorgiaTech. Previous Intern @Google & @nvidia. All opinions are my own.

🤯 Merging many finetuned LLMs into one model, effectively? Introducing Functional Dual Anchor (FDA), a new framework for model merging. 🚀 Current merging works poorly due to the underlying parameter conflicts. FDA shifts knowledge integration to the input-representation space…

Besteuler's tweet image. 🤯 Merging many finetuned LLMs into one model, effectively? Introducing Functional Dual Anchor (FDA), a new framework for model merging.

🚀 Current merging works poorly due to the underlying parameter conflicts. FDA shifts knowledge integration to the input-representation space…

The physics prior matters in molecular structures. We model potential energy between molecules for drug design. This happens to have a coincident yet interesting connection to my past work, hyperspherical energy (arxiv.org/abs/1805.09298), which considers potential energy between…

Besteuler's tweet image. The physics prior matters in molecular structures. We model potential energy between molecules for drug design. This happens to have a coincident yet interesting connection to my past work, hyperspherical energy (arxiv.org/abs/1805.09298), which considers potential energy between…

Weiyang Liu reposted

SMPL is 10 years old and has done what we hoped — it changed the way the field estimates and models 3D humans and their motion. I’m delighted that the original team has been recognized today at @ICCVConference with the Mark Everingham Prize. The prize is given to individuals or…

Michael_J_Black's tweet image. SMPL is 10 years old and has done what we hoped — it changed the way the field estimates and models 3D humans and their motion. I’m delighted that the original team has been recognized today at @ICCVConference
with the Mark Everingham Prize. 

The prize is given to individuals or…

This is almost a year-long project and led by @ItsTheZhen. My biggest takeaway is that physical simulation is very effective as a reward signal, and this efficient verification is crucial for unlocking LLMs’ design novelty. This conclusion is actually aligned with our previous…

Can LLMs design real machines — from 🚗 cars to 🏹 catapults? Can they engineer through both 🧠 agentic workflows and 🌀 reinforcement learning (RL) — learning from physical simulation instead of text alone? We treat machine design as “machine code writing”, where LLMs assemble…

ItsTheZhen's tweet image. Can LLMs design real machines — from 🚗 cars to 🏹 catapults?
Can they engineer through both 🧠 agentic workflows and 🌀 reinforcement learning (RL) — learning from physical simulation instead of text alone?

We treat machine design as “machine code writing”, where LLMs assemble…


Weiyang Liu reposted

Prof. Chen Ning Yang, a world-renowned physicist, Nobel Laureate in Physics, Academician of the Chinese Academy of Sciences, Professor at Tsinghua University, and Honorary Director of the Institute for Advanced Study at Tsinghua University, passed away in Beijing due to illness…

Tsinghua_Uni's tweet image. Prof. Chen Ning Yang, a world-renowned physicist, Nobel Laureate in Physics, Academician of the Chinese Academy of Sciences, Professor at Tsinghua University, and Honorary Director of the Institute for Advanced Study at Tsinghua University, passed away in Beijing due to illness…

🤖 Can LLMs learn to create? Introducing "Agentic Design of Compositional Machines" — a new frontier where AI builds functional machines from standardized parts. We present BesiegeField, a simulation testbed to benchmark LLMs on tasks like building cars & catapults. Key…

Besteuler's tweet image. 🤖 Can LLMs learn to create? Introducing "Agentic Design of Compositional Machines" — a new frontier where AI builds functional machines from standardized parts.

We present BesiegeField, a simulation testbed to benchmark LLMs on tasks like building cars & catapults. Key…

Weiyang Liu reposted

🚀 Excited to share our new paper: "SimKO: Simple Pass@K Policy Optimization"! SimKO is a new algorithm for effectively boosts pass@K performance on math & logic tasks without sacrificing pass@1. spherelab.ai/simko (1/n)

RuotianPeng's tweet image. 🚀 Excited to share our new paper: "SimKO: Simple Pass@K Policy Optimization"!  

SimKO is a new algorithm for effectively boosts pass@K performance on math & logic tasks without sacrificing pass@1.  

spherelab.ai/simko  
(1/n)

Weiyang Liu reposted

TL;DR: Meet BesiegeField—a playground where LLMs build, test, and refine machines from standard parts in real time. We tested agentic workflows and RLVR with top LLMs: even the strongest still show limits in compositional machine design. 🔗 besiegefield.github.io 🧵 below

ItsTheZhen's tweet image. TL;DR: Meet BesiegeField—a playground where LLMs build, test, and refine machines from standard parts in real time.

We tested agentic workflows and RLVR with top LLMs: even the strongest still show limits in compositional machine design.

🔗 besiegefield.github.io
🧵 below

Human history is marked by the machines we created: from the Antikythera mechanism of ancient Greece, to the imaginations of the Renaissance, to the engines of the steam era. We wonder: can LLMs, like humans, build sophisticated machines to achieve purposeful functionality?



This is a wonderful collaboration with @ItsTheZhen and Wenqian. I’ve long been curious whether large language models truly possess creativity -- the ability to build something genuinely novel. This project represents our first step toward answering that question. It also aligns…

Human history is marked by the machines we created: from the Antikythera mechanism of ancient Greece, to the imaginations of the Renaissance, to the engines of the steam era. We wonder: can LLMs, like humans, build sophisticated machines to achieve purposeful functionality?



🚀 Glad to introduce SimKO (Simple Pass@K Optimization) Current GRPO-based methods overfit to safe responses -- great Pass@1, poor Pass@K. 🔍 We find this stems from probability over-concentration: the model collapses onto its top-1 token, losing exploration. This appears to be…

Besteuler's tweet image. 🚀 Glad to introduce SimKO (Simple Pass@K Optimization)

Current GRPO-based methods overfit to safe responses -- great Pass@1, poor Pass@K.
🔍 We find this stems from probability over-concentration: the model collapses onto its top-1 token, losing exploration. This appears to be…

Loading...

Something went wrong.


Something went wrong.