Welcome :)

Hello there! I’m Ke Yang (杨可), currently a second-year Ph.D. at UIUC under the guidance of Professor ChengXiang Zhai. I obtain my bachelor’s degree from Tsinghua University. I interned with Professor Heng Ji’s group at UIUC in 2022 summer, and at Amazon AWS in 2024 summer.

I work on intelligent agents, language models, graph neural networks, and multimodality foundation models (of top interest). I’m also keen on NLP for societal benefit.

During the winter break of 2022, I collaborated with two of my undergrad classmates and created Zempath, an online social platform incorporated with our trained chatbots with distinctive personalities.

Selected Publications

arXiv 2024
yang2024tinyhelenscurriculumtrainingevaluating

TinyHelen's First Curriculum: Training and Evaluating Tiny Language Models in a Simpler Language Environment

Ke Yang, Volodymyr Kindratenko, ChengXiang Zhai

We train and evaluate tiny language models using a text dataset with simplified vocabularies and linguistic structures, mimicking how children learn language through simplified environments as part of their initial curriculum.

ICLR 2025
yang2024agentoccamsimplestrongbaseline

AgentOccam: A Simple Yet Strong Baseline for LLM-Based Web Agents

Ke Yang, Yao Liu, Sapana Chaudhary, Rasool Fakoor, Pratik Chaudhari, George Karypis, Huzefa Rangwala

AgentOccam surpasses the previous state-of-the-art and concurrent LLM-based web agent with its observation and action space alignment. We achieve this without using in-context examples, new agent roles, online feedback or search strategies.

NeurIPS 2024 D&B Track
2024prejudice

Bias and Volatility: A Statistical Framework for Evaluating Large Language Model's Stereotypes and the Associated Generation Inconsistency

Yiran Liu*, Ke Yang*, Zehan Qi, Xiao Liu, Yang Yu, ChengXiang Zhai (* indicates equal contributions)

Bias-Volatility Framework measures discrimination in models by considering both their consistently biased preference and preference variation across contexts.

ICLR Workshop 2024
yang2024llm

If LLM Is the Wizard, Then Code Is the Wand: A Survey on How Code Empowers Large Language Models to Serve as Intelligent Agents

Ke Yang*, Jiateng Liu*, John Wu, Chaoqi Yang, Yi R. Fung, Sha Li, Zixuan Huang, Xu Cao, Xingyao Wang, Yiquan Wang, Heng Ji, ChengXiang Zhai (* indicates equal contributions)

The Wizard survey explores the synergy between code and large language models (LLMs), highlighting how code empowers LLMs and benefits LLM when they serve as intelligent agents. We emphasized code’s readability, symbolic abstraction, and graph structure, presenting it as a valuable component in LLMs’ training corpus.

AAAI 23
yang2022adept

ADEPT: A DEbiasing PrompT Framework

Ke Yang, Charles Yu, Yi Fung, Manling Li, Heng Ji

ADEPT introduces a novel debaising loss function based on counterfactual bias and manifold learning insights. “Prompt” here refers to prompt-tuning (peft) rather than prompt-engineering.

Icon Zempath

In the promotional video for Zempath, we unveil our driving inspirations and fundamental principles. We showcase the seamless user experience of engaging in chats, posting either anonymously or under one’s real name, indulging in conversations with our personalized chatbots, and forging new connections with like-minded individuals. Let’s delve into a snippet from this captivating video:

Zempath

Miscellaneous

I am an amateur novelist, painter, and photographer. I take photos of cats, my sister, grandparents, friends, campus, etc., in my spare time.