4 citations · 4 across the 14 of their papers we have counts for
14 papers
Beyond Surface Alignment: Grounding the Dynamics of Situational Understanding and Generative Control in LLMs
Chenghao Yang
The current alignment tuning paradigm for Large Language Models (LLMs) prioritizes surface-level behaviors -- fluency, safety, and tonal consistency. While effective for casual cha…
Value-Aware Stochastic KV Cache Eviction for Reasoning Models
Ting-Yun Chang, Harvey Yiyun Fu, Deqing Fu +3
Reasoning models improve accuracy through extended chains of thought, but their long outputs create a memory and compute bottleneck. KV cache eviction methods reduce this cost by e…
The Chameleon's Limit: Investigating Persona Collapse and Homogenization in Large Language Models
Yunze Xiao, Vivienne J. Zhang, Chenghao Yang +3
Large language models (LLMs) have powered multi-agent simulations such as survey respondents or application users. However, LLMs lack the population-level diversity required by the…
When Agents Look the Same: Quantifying Distillation-Induced Similarity in Tool-Use Behaviors
Chenghao Yang, Yuning Zhang, Zhoufutu Wen +4
Model distillation is a primary driver behind the rapid progress of LLM agents, yet it often leads to behavioral homogenization. Many emerging agents share nearly identical reasoni…
WorldTravel: A Realistic Multimodal Travel-Planning Benchmark with Tightly Coupled Constraints
Zexuan Wang, Chenghao Yang, Yingqi Que +18
Real-world autonomous planning requires coordinating tightly coupled constraints where a single decision dictates the feasibility of all subsequent actions. However, existing bench…
MME-CC: A Challenging Multi-Modal Evaluation Benchmark of Cognitive Capacity
Kaiyuan Zhang, Chenghao Yang, Zhoufutu Wen +19
As reasoning models scale rapidly, the essential role of multimodality in human cognition has come into sharp relief, driving a growing need to probe vision-centric cognitive behav…
DiscoX: Benchmarking Discourse-Level Translation task in Expert Domains
Xiying Zhao, Zhoufutu Wen, Zhixuan Chen +20
The evaluation of discourse-level translation in expert domains remains inadequate, despite its centrality to knowledge dissemination and cross-lingual scholarly communication. Whi…
Optimizing Diversity and Quality through Base-Aligned Model Collaboration
Yichen Wang, Chenghao Yang, Tenghao Huang +3
Alignment has greatly improved large language models (LLMs)' output quality at the cost of diversity, yielding highly similar outputs across generations, especially in open-ended g…
FinSearchComp: Towards a Realistic, Expert-Level Evaluation of Financial Search and Reasoning
Liang Hu, Jianpeng Jiao, Jiashuo Liu +20
Search has emerged as core infrastructure for LLM-based agents and is widely viewed as critical on the path toward more general intelligence. Finance is a particularly demanding pr…
MARS-Bench: A Multi-turn Athletic Real-world Scenario Benchmark for Dialogue Evaluation
Chenghao Yang, Yinbo Luo, Zhoufutu Wen +8
Large Language Models (\textbf{LLMs}), e.g. ChatGPT, have been widely adopted in real-world dialogue applications. However, LLMs' robustness, especially in handling long complex di…
CryptoX : Compositional Reasoning Evaluation of Large Language Models
Jiajun Shi, Chaoren Wei, Liqun Yang +7
The compositional reasoning capacity has long been regarded as critical to the generalization and intelligence emergence of large language models LLMs. However, despite numerous re…
Hello Again! LLM-powered Personalized Agent for Long-term Dialogue
Hao Li, Chenghao Yang, An Zhang +3
Open-domain dialogue systems have seen remarkable advancements with the development of large language models (LLMs). Nonetheless, most existing dialogue systems predominantly focus…
Equipping Transformer with Random-Access Reading for Long-Context Understanding
Chenghao Yang, Zi Yang, Nan Hua
Long-context modeling presents a significant challenge for transformer-based large language models (LLMs) due to the quadratic complexity of the self-attention mechanism and issues…
Can You Follow Me? Testing Situational Understanding in ChatGPT
Chenghao Yang, Allyson Ettinger
Understanding sentence meanings and updating information states appropriately across time -- what we call "situational understanding" (SU) -- is a critical ability for human-like A…