Papers by Donghai Zhang
MARIO-0.5B: A Multi-Agent Lightweight Model for Real-Time Open Information Extraction in Low-Resource Settings (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Large language models have shown remarkable capabilities in open information extraction, but their resource requirements often restrict their deployment in resource-constrained industrial settings. |
| Approach: | They introduce an ultra-lightweight large language model trained on instruction-based samples in Chinese, English, Korean, and Russian. |
| Outcome: | The proposed model outperforms large-scale models with up to 70B parameters, reducing computational resources by 140x and delivering 11x faster response times. |
Modeling Sentiment Association in Discourse for Humor Recognition (P18-2)
Copied to clipboard
| Challenge: | Existing work on sentiment information is limited to the number of emotional words. |
| Approach: | They propose to model sentiment association between discourse units to indicate how punchline breaks expectation of setup. |
| Outcome: | The proposed model shows that discourse relation, sentiment conflict and sentiment transition are effective indicators for humor recognition. |
PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference (2025.acl-long)
Copied to clipboard
Jiaming Ji, Donghai Hong, Borong Zhang, Boyuan Chen, Josef Dai, Boren Zheng, Tianyi Alex Qiu, Jiayi Zhou, Kaile Wang, Boxun Li, Sirui Han, Yike Guo, Yaodong Yang
| Challenge: | Using large-scale annotation data, large language models can generate noise, errors and biases, leading to unexpected behaviours. |
| Approach: | They propose a dataset to promote safety alignment in large language models . they separate helpfulness and harmlessness annotations for question-answering pairs . |
| Outcome: | The proposed dataset provides 44.6k prompts and 265k question-answer pairs with safety meta-labels for 19 harm categories and three severity levels, with answers generated by Llama-family models. |
Exploiting Syntactic Structures for Humor Recognition (C18-1)
Copied to clipboard
| Challenge: | Using syntactic structure features, we find humor recognition is a kind of style . |
| Approach: | They propose to exploit syntactic structure features to enhance humor recognition . they find syntastic structure features consistently correlate with humor . |
| Outcome: | The proposed method achieves significant improvements compared with baselines. |