Papers by Shijie Lu
Academics Can Contribute to Domain-Specialized Language Models (2024.emnlp-main)
Copied to clipboard
Mark Dredze, Genta Winata, Prabhanjan Kambadur, Shijie Wu, Ozan Irsoy, Steven Lu, Vadim Dabravolski, David Rosenberg, Sebastian Gehrmann
| Challenge: | Commercially available models dominate academic leaderboards, focusing on creating and adapting general-purpose models . however, general- purpose models often underperform in specialized domains, and domain-specific models yield superior results. |
| Approach: | They advocate for a renewed focus on developing and evaluating domain- and task-specific models . they advocate for an adapted or adapted model that can be used to improve academic leaderboard standings . |
| Outcome: | The proposed model can do well on professional and linguistic examinations, college-level knowledge questions, and collections of reasoning tasks. |
ToW: Thoughts of Words Improve Reasoning in Large Language Models (2025.naacl-long)
Copied to clipboard
Zhikun Xu, Ming Shen, Jacob Dineen, Zhaonan Li, Xiao Ye, Shijie Lu, Aswin Rrv, Chitta Baral, Ben Zhou
| Challenge: | Unlike other data augmentation methods, thoughts of words (TOW) views next-word prediction as a core reasoning task and injects fine-grained thoughts into pre-training texts. |
| Approach: | They propose a training-time data-augmentation method called thoughts of words (TOW) that injects fine-grained thoughts directly into a next-word prediction task and teaches the model to understand how the observed next word is related to previous contexts. |
| Outcome: | The proposed method reduces model hallucination by 10% and improves reasoning performance by 7% to 9% on average. |
MixCE: Training Autoregressive Language Models by Mixing Forward and Reverse Cross-Entropies (2023.acl-long)
Copied to clipboard
| Challenge: | Autoregressive language models are trained by minimizing the cross-entropy of the model distribution Q relative to the data distribution P. However, these systems still struggle in many openended generation settings, where they are asked to produce a long text following a short prompt. |
| Approach: | They propose to combine forward and reverse cross-entropy to train autoregressive language models by minimizing the cross-Entropy of the model distribution Q relative to the data distribution P. |
| Outcome: | The proposed model overgeneralizes and produces non-human-like text without complex decoding strategies. |
AutoSDT: Scaling Data-Driven Discovery Tasks Toward Open Co-Scientists (2025.emnlp-main)
Copied to clipboard
Yifei Li, Hanane Nour Moussa, Ziru Chen, Shijie Chen, Botao Yu, Mingyi Xue, Benjamin Burns, Tzu-Yao Chiu, Vishal Dey, Zitong Lu, Chen Wei, Qianheng Zhang, Tianyu Zhang, Song Gao, Xuhui Huang, Xia Ning, Nesreen K. Ahmed, Ali Payani, Huan Sun
| Challenge: | AutoSDT-5K is the only automatically collected and the largest open dataset for data-driven scientific discovery. |
| Approach: | They propose an automatic pipeline that collects high-quality coding tasks in real-world data-driven discovery workflows. |
| Outcome: | The proposed pipeline synthesizes accurate tasks and tasks from a dataset of 5,404 tasks covering four scientific disciplines and 756 Python packages. |
QA‐LIGN: Aligning LLMs through Constitutionally Decomposed QA (2025.findings-emnlp)
Copied to clipboard
Jacob Dineen, Aswin Rrv, Qin Liu, Zhikun Xu, Xiao Ye, Ming Shen, Zhaonan Li, Shijie Lu, Chitta Baral, Muhao Chen, Ben Zhou
| Challenge: | QA-LIGN decomposes monolithic rewards into interpretable principle-specific evaluations . scalar rewards obscure which objectives drive the training signal . |
| Approach: | a new method decomposes monolithic rewards into interpretable principle-specific evaluations . QA-LIGN reduces attack success rates by up to 68.7% while maintaining a 0.67% false refusal rate . |
| Outcome: | QA-LIGN reduces attack success rates by up to 68.7% while maintaining a 0.67% false refusal rate . the results outperform DPO and GRPO with state-of-the-art reward models given equivalent training . |