Log inSign up
Yuji Zhang
124 posts
user avatar
Yuji Zhang
@Yuji_Zhang_NLP
Postdoc@UIUC, advised by Prof. Heng Ji @hengjinlp and Prof. Chengxiang Zhai. Robust and trustworthy foundation model. Knowledge. Reasoning. Hallucination.
Urbana, IL
Joined August 2020
448
Following
769
Followers
  • Pinned
    user avatar
    Yuji Zhang
    @Yuji_Zhang_NLP
    Mar 1, 2025
    🔍New findings of knowledge overshadowing! Why do LLMs hallucinate over all true training data? 🤔Can we predict hallucinations even before model training or inference? 🚀Check out our new preprint: [arxiv.org/pdf/2502.16143] The Law of Knowledge Overshadowing: Towards
    Image
    Image
    Image
    Image
    27K
  • user avatar
    Yuji Zhang
    @Yuji_Zhang_NLP
    Jul 29, 2024
    🔍 New Preprint! Why do LLMs generate hallucinations even when trained on all truths? 🤔 Check out our paper [arxiv.org/abs/2407.08039] 💡 We find that universally, data imbalance causes LLMs to over-generalize popular knowledge and produce amalgamated hallucinations. 📊
    Image
    Image
    Image
    Image
    77K
  • user avatar
    Yuji Zhang
    @Yuji_Zhang_NLP
    Jul 29, 2024
    💡Knowledge overshadowing caused by data imbalance makes LLMs hallucinate even when trained on all true statements🧐? arxiv.org/abs/2407.08039 💡Hallucination is generalization?!😲 How can we balance between hallucinations and intelligence brought by generalization?
    Image
    Image
    Image
    Image
    8.7K
  • user avatar
    Yuji Zhang
    @Yuji_Zhang_NLP
    Jul 29, 2024
    Replying to @Yuji_Zhang_NLP
    Deep appreciation to my co-authors for their amazing support and great suggestions!🤩😊 @ZoeyLi20 @JiatengLiu PengfeiYu @YiFung10 @ManlingLi_ @hengjinlp
    1.9K
  • user avatar
    Yuji Zhang
    @Yuji_Zhang_NLP
    Oct 8, 2023
    I am glad our paper has been accepted by EMNLP 2023. 🎉 VIBE: Topic-Driven Temporal Adaptation for Twitter Classification. Yuji Zhang, Jing Li, Wenjie Li. VIBE explores how to adapt language models to the future in the continuously changing environments.
    411
  • user avatar
    Yuji Zhang
    @Yuji_Zhang_NLP
    Jul 29, 2024
    Replying to @Yuji_Zhang_NLP
    Big thanks to my co-authors for their awesome support and great suggestions!!!@ZoeyLi20 @JiatengLiu @YiFung10 @ManlingLi_ @hengjinlp
    547
  • user avatar
    Yuji Zhang
    @Yuji_Zhang_NLP
    Jul 29, 2024
    Replying to @zoewangai
    Thanks! Truly the imbalanced word patterns (and knowledge) are ubiquitous in training data, thus it would both challenging and meaningful for us to dive deeper into better training distribution, strategies, and architectures :)
    413
  • user avatar
    Yuji Zhang
    @Yuji_Zhang_NLP
    Jul 29, 2024
    Replying to @flesheatingemu
    The "actual AI hallucination" may indicate higher-level intelligence in the future. Look forward to exploring it!
    329
  • user avatar
    Yuji Zhang
    @Yuji_Zhang_NLP
    Jul 29, 2024
    Replying to @hbouammar
    Thanks! Indeed knowledge overshadowing universally exists in various domains, including the group-bias domain. To mitigate the prior bias introduced by training data, we utilize the inference-time SCD method for broader applications, and we believe RL methods could also help in
    730
  • user avatar
    Yuji Zhang
    @Yuji_Zhang_NLP
    Oct 8, 2023
    Replying to @Yuji_Zhang_NLP
    #EMNLP2023 #NLP #timetravel
    315
  • user avatar
    Yuji Zhang
    @Yuji_Zhang_NLP
    Oct 20, 2023
    Replying to @Yuji_Zhang_NLP
    Our paper is now available:
    arXiv logo
    arxiv.org
    VIBE: Topic-Driven Temporal Adaptation for Twitter Classification
    Language features are evolving in real-world social media, resulting in the deteriorating performance of text classification in dynamics. To address this challenge, we study temporal adaptation,...
    286
  • user avatar
    Yuji Zhang
    @Yuji_Zhang_NLP
    Jul 30, 2024
    Replying to @TonyCheng990417
    Thanks!🤓 We have experimented with 160m-7b models on fine-tuning tasks and 1b-13b models on inference-time ones (Table 4, 5). Interestingly, we observed the reverse scaling tendency with the model sizes, showing that knowledge overshadowing exacerbates with the increasing model
    201
  • user avatar
    Yuji Zhang
    @Yuji_Zhang_NLP
    Jul 30, 2024
    Replying to @flesheatingemu
    Thanks for this interesting question! That would a challenging but very significant direction to explore considering the existing fallacies of current architectures. If we want to achieve that, we should resort to human-like architectures and strategies that encourage more
    56
  • user avatar
    Yuji Zhang
    @Yuji_Zhang_NLP
    Dec 6, 2023
    Replying to @May_F1_ and @taoyds
    I can not wait🤭 Hope to see you soon
    56

New to X?

Sign up now to get your own personalized timeline!

Create account

By signing up, you agree to the Terms of Service and Privacy Policy, including Cookie Use.

Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
Don't miss what's happening
People on X are the first to know.
Log inSign up
Advertisement
Advertisement