Jiyoung Lee

[Jiyoung Lee @jiyounglee0523]

Curious how your sentence reads in Scottish English, or as written by a French A-level learner? Give it a try — and feel free to share your favorite transformation in the comments 👇

📄 Paper: https://t.co/zShe6L2z1F
💻 Code: https://t.co/qFxZU3Kgo8

[Jiyoung Lee @jiyounglee0523]

Did you know LLMs can get noticeably worse at the same task when the question isn't written in Standard American English?

Following up on our NeurIPS 2025 paper Trans-EnV — we just launched a live demo you can try right now! 🎉
🎛️ Try it here: https://t.co/ueewqYa4pe

[Jiyoung Lee @jiyounglee0523]

What you can do in the demo:
🔹 Type any English sentence
🔹 Pick one of 38 English varieties — 18 dialects (AAVE, Scottish, Irish, Australian, …) or 20 ESL learner varieties (10 native languages × CEFR proficiency levels)
🔹 Watch each linguistic rule applied step by step

[Jiyoung Lee @jiyounglee0523]

For more details, check out our paper and project page! This work was done in collaboration with Minwoo Kim, @seungho_k, Junghwan Kim, Seunghyun Won, @hwaran_lee, and Edward Choi.

paper: https://t.co/dMfYfPM8Fk
project page: https://t.co/ib1pgJXz6n

[Jiyoung Lee @jiyounglee0523]

🤔 Have you ever wondered how much LMs are aligned to YOUR COUNTRY?

We constructed KorNAT (Korean National Alignment Test), the first benchmark that measures national alignment between LMs and South Korea! (1/n)

paper: https://t.co/dMfYfPM8Fk

[Jiyoung Lee @jiyounglee0523]

Our dataset creation process is meticulously designed and was refined through multiple rounds of human reviews. KorNAT passed both qualitative and quantitative assessments by a government-affiliated organization dedicated to evaluating dataset quality. (9/n)

[Jia-Bin Huang @jbhuang0604]

How to write the Introduction?

As a junior student, writing the introduction of a research paper is arguably the most daunting part of paper writing. 😱

Here is a simple template I find useful:

3 Figures 🖼️ + 5 Questions 🤔

[chilconference @CHILconference]

Can we transfer predictive model to different EHR systems without code mapping process?
Our text-based code embedding approach unifies code systems & enables transfer and pooled learning in heterogenous EHR systems. https://t.co/c5jEqw1bdx
@jiyounglee0523 @Jwooooo5 @Wes25Price

[Jiyoung Lee @jiyounglee0523]

@karthikabinav Will there also be research interns in other area, such as multi-modality?

[NVIDIA AI @NVIDIAAI]

One main shortcoming of generative language models is their lack of controllability and consistency with real-world facts. Join us at GTC to watch NVIDIA researchers propose a new generation framework, Megatron-CNTRL, to address these limitations. #GTC20 https://t.co/xrzw0JjRpk

[Danijar Hafner @danijarh]

World models are the future and the future is now! 🌎🚀

Proud to share DreamerV2, the first agent that achieves human-level Atari performance by learning behaviors purely within a separately trained world model.

Paper: https://t.co/deRLWwVZ8Q

Thread 👇