Zhuang Li
@Zhuang_Li_NLP
PhD, Prev Research Fellow @MonashUni | Prev SDE @Microsoft | NLP | Contribution @BigCodeProject | Lecturer @RMIT | ARR Area Chair
You might like
Modern LLMs create a dilemma: powerful models are costly, cheap models are unreliable. xRouter is a learned routing system trained via reinforcement learning with explicit cost-performance trade-offs. 🤖 Paper: bit.ly/49e0v3E Key results: substantial cost reductions…
Happy to announce that my #EMNLP2025 paper Humanizing Machines: Rethinking LLM Anthropomorphism Through a Multi-Level Framework of Design have finally made it to arxiv! This work REDEFINES anthropomorphism in LLM!! arxiv.org/abs/2508.17573…
✨We are thrilled to announce that over 3200 papers have been accepted to #EMNLP2025 ✨ This includes over 1800 main conference papers and over 1400 papers in findings! Congratulations to all authors!! 🎉🎉🎉
Tbh I'm happy to see ChatGPT’s downloads reaching major social media apps combined! Beyond work (startup & coding), it's been a life-changer for me: - Cured my dizziness. Two doctors couldn't help, but ChatGPT suggested electrolyte water. It worked! - fixed my e-bike myself. New…
🎉 Our paper SCAR (Style Consistency-Aware Ranking) is accepted to ACL 2025 as a poster presentation! See you in Vienna! 🚀 In the best case, using just 0.7% of the original data, SCAR enables OLMo-7B to outperform its full-data fine-tuned counterpart and consistently surpasses…
BESSTIE is a new dataset for sentiment and sarcasm classification of Australian, Indian and British English. Do check it out! #ACL2025
Keep those reviews coming in! We are currently up to 78% of papers with three reviews submitted. ▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓▓░░░░░░░ 78%
📢 ACL 2025 | LAZYREVIEW (Meta Score: 5/5!) Is peer review always fair and constructive? We present LAZYREVIEW: the first dataset that exposes lazy thinking in NLP reviews: vague rejections like “Not novel enough” or “Only tested on German” without solid justification. 💡 What…
6 papers accepted to ACL 2025 main track! Work mainly focuses on LLM safety, global diversity, and low-resource NLP. Thanks to all amazing collaborators—see you in Vienna! More details on papers soon.
🌏 New position paper analyzes 2,000+ multilingual LLM benchmarks from 148 countries (2021-2024) and reveals key challenges in multilingual AI evaluation. 🌺
🥳🥳🥳New dataset: huggingface.co/datasets/mingh… We recently received some free compute, so we created a synthetic dataset of 10M realistic personas using meta-llama/Llama-3.3-70B-Instruct & Qwen/Qwen2.5-72B-Instruct. Each persona includes features like name, DOB, personality, and…
Can LLMs spot signs of self-harm in multilingual, culturally nuanced contexts? JiraiBench is our new benchmark focusing on the Jirai (地雷系) subculture in Japan & China—offering a hard testbed for LLMs on detection + reasoning. arxiv.org/pdf/2503.21679
Hiring Lecturer in NLP: jobs.surrey.ac.uk/vacancy.aspx?r…. Lectureship position aligned with Nature Inspired Computing and Engineering (NICE) research group (lnkd.in/ew6xpiJm) within the CS. Happy to be contacted for discussing our research/strengths. #nlproc
Just finished wrapping up 6 submissions in ACL. Being both an AC and a reviewer, I guess my review load will be exploding this time…
Hao's PhD research in audio-safety red teaming of LLMs has now extended into a new exciting direction in his latest #NAACL2025 paper. In his recent work "Audio Is the Achilles' Heel: Red Teaming Audio Large Multimodal Models" we ask the following questions: (1) Do text-only LLMs…
Speech (or audio to be more specific) related safety is literally unexplored beyond content. If the focus is only placed on safeguarding "what" is being said but not "how" it is sad or in "which context" it is said, then we are left with very weak safety measures for speech. A…
Thrilled to share that our paper “CultureInstruct: Curating Large-Scale Multi-Cultural Instructions” has been accepted to #NAACL2025 main! 🎉 It introduces a novel data synthesis method for reducing cultural bias in LLMs. Stay tuned-preprint coming soon on arXiv! 🚀 #AI #NLP
United States Trends
- 1. #Worlds2025 46.1K posts
- 2. Doran 18.6K posts
- 3. Good Sunday 47.4K posts
- 4. #T1WIN 29.2K posts
- 5. Faker 32K posts
- 6. Sam Houston 1,623 posts
- 7. Oregon State 4,631 posts
- 8. #T1fighting 3,498 posts
- 9. Option 2 4,328 posts
- 10. Vergil 8,771 posts
- 11. Boots 29.3K posts
- 12. #Toonami 2,731 posts
- 13. Keria 11.2K posts
- 14. Hyan 1,494 posts
- 15. Louisville 14.3K posts
- 16. UCLA 7,765 posts
- 17. #FlyTogether 2,132 posts
- 18. Nikki Glaser 1,039 posts
- 19. OutKast 22.2K posts
- 20. PETE DAVIDSON N/A
Something went wrong.
Something went wrong.