Papers by Chiyu Zhang
The Skipped Beat: A Study of Sociopragmatic Understanding in LLMs for 64 Languages (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing instruction tuned large language models (LLMs) struggle to understand cross-lingual sociopragmatic meaning (SM) lack of comprehensive investigation into their ability to understand SM is partly due to SM not being adequately represented in any of the existing benchmarks. |
| Approach: | They evaluate the performance of instruction tuned large language models (LLMs) on a multilingual benchmark specifically designed for SM understanding. |
| Outcome: | The proposed benchmark outperforms instruction tuned large language models on a wide range of tasks but falls behind task-specific finetuned models. |
LaMini-LM: A Diverse Herd of Distilled Models from Large-Scale Instructions (2024.eacl-long)
Copied to clipboard
| Challenge: | Large language models with instruction tuning are resource-intensive . a recent study suggests that the performance of LLMs scales proportionally with the size of the model. |
| Approach: | They propose to distill knowledge from instruction-tuned LLMs into much smaller ones . they develop a large set of 2.58M instructions based on existing and newly-generated instructions . |
| Outcome: | The proposed models are comparable to strong baselines while being much smaller in size. |
Music Audio-Visual Question Answering Requires Specialized Multimodal Designs (2026.findings-acl)
Copied to clipboard
Wenhao You, Xingjian Diao, Wenjun Huang, Chunhui Zhang, Keyi Kong, Weiyi Wu, Chiyu Ma, Zhongyu Ouyang, Tingxuan Wu, Ming Cheng, Soroush Vosoughi, Jiang Gui
| Challenge: | Music audio-visual question answering presents unique challenges with dense audio-visual content, intricate temporal dynamics, and the need for domain-specific knowledge. |
| Approach: | They analyze Music AVQA datasets and analyze their results to identify key design patterns . they propose concrete future directions for incorporating musical priors . |
| Outcome: | The proposed architectures are critical for success in Music AVQA, the authors argue . they suggest concrete future directions for incorporating musical priors . |
Contrastive Learning of Sociopragmatic Meaning in Social Media (2023.findings-acl)
Copied to clipboard
| Challenge: | Recent progress in representation and contrastive learning in NLP has not considered the class of sociopragmatic meaning (i.e., meaning in interaction within different language communities). |
| Approach: | They propose a framework for learning task-agnostic representations transferable to a wide range of sociopragmatic tasks. |
| Outcome: | The proposed framework outperforms other contrastive learning frameworks for both in-domain and out-of-domain data, across both the general and few-shot settings. |
Distilling Text Style Transfer With Self-Explanation From LLMs (2024.naacl-srw)
Copied to clipboard
| Challenge: | Text Style Transfer (TST) aims to alter the style of text while preserving its core content. |
| Approach: | They propose a framework that leverages large language models alongside chain-of-thought prompting to facilitate TST. |
| Outcome: | The proposed framework surpasses supervised fine-tuning and knowledge distillation methods in low-resource settings. |
Judging with Many Minds: Do More Perspectives Mean Less Prejudice? On Bias Amplification and Resistance in Multi-Agent Based LLM-as-Judge (2025.findings-emnlp)
Copied to clipboard
Chiyu Ma, Enpei Zhang, Yilun Zhao, Wenjun Liu, Yaning Jia, Peijun Qing, Lin Shi, Arman Cohan, Yujun Yan, Soroush Vosoughi
| Challenge: | LLM-as-Judge frameworks provide scalable alternative to human evaluation . but the question of how intrinsic biases manifest in these settings remains unexplored . |
| Approach: | They conduct systematic analysis of four bias types in multi-agent LLM-as-Judge frameworks . they find debate framework amplifies biases sharply after initial debate . |
| Outcome: | The proposed frameworks amplify biases after debate and show they are stronger in meta-judge scenarios. |
SoundMind: RL-Incentivized Logic Reasoning for Audio-Language Models (2025.emnlp-main)
Copied to clipboard
Xingjian Diao, Chunhui Zhang, Keyi Kong, Weiyi Wu, Chiyu Ma, Zhongyu Ouyang, Peijun Qing, Soroush Vosoughi, Jiang Gui
| Challenge: | Recent large language models have demonstrated impressive reasoning abilities, but their extension to the audio modality remains underexplored. |
| Approach: | They propose a rule-based reinforcement learning algorithm to equip LALMs with robust reasoning capabilities. |
| Outcome: | The proposed algorithm improves on the SoundMind benchmark. |
Dynamics of Instruction Fine-Tuning for Chinese Large Language Models (2025.coling-main)
Copied to clipboard
| Challenge: | Instruction tuning is a burgeoning method to elicit the general intelligence of Large Language Models. |
| Approach: | They investigate the effects of data quantity, model size, and data construction methods on instruction tuning for Chinese LLMs. |
| Outcome: | The proposed model includes over 40,000 high-quality instruction instances covering ten underlying abilities. |
Toward Micro-Dialect Identification in Diaglossic and Code-Switched Environments (2020.emnlp-main)
Copied to clipboard
| Challenge: | Existing work on dialect prediction is limited to coarse-grained varieties . a new language model, MARBERT, can predict micro-dialects with 9.9% F1, 76 better than a majority class baseline. |
| Approach: | They propose a new task of Micro-Dialect Identification (MDI) that can predict a fine-grained variety given a single message. |
| Outcome: | The proposed model predicts micro-dialects with 9.9% F1, 76 better than a majority class baseline. |
Demystifying Instruction Mixing for Fine-tuning Large Language Models (2024.acl-srw)
Copied to clipboard
| Challenge: | Instruction tuning is effective for aligning large language models with human instructions, but the procedure to optimizing the mixing of instruction datasets is still unclear. |
| Approach: | They categorize instructions into three primary types: NLP downstream tasks, coding, and general chat. |
| Outcome: | The proposed method improves performance of large language models (LLMs) but it is difficult to combine different instruction datasets to optimize overall performance. |
What Makes a Good Curriculum? Disentangling the Effects of Data Ordering on LLM Mathematical Reasoning (2026.acl-long)
Copied to clipboard
| Challenge: | Curriculum learning (CL) orders data corpus by difficulty, but prior work employs disparate difficulty metrics and training setups. |
| Approach: | They propose a framework that decomposes curriculum difficulty into five dimensions: Problem Difficulty, Model Surprisal, Confidence Margin, Predictive Uncertainty and Decision Variability. |
| Outcome: | The proposed framework decomposes curriculum difficulty into five dimensions . the results show that no curriculum strategy dominates universally . |