Working at the Singapore office is very appealing! Really miss all the food there 😂

When I was a PhD student, my advisor taught me what good management of technical talent looks like. We used to meet on a weekly basis, and towards the end of that week-long period of banging my head against the wall on an ambiguous problem, my enthusiasm and drive to dig deeper would have dampened quite significantly. Some weeks I had promising results to show, some weeks I just had results. Some weeks, I brought in issues and insights and asked for direction. Some weeks, when my experiments failed to validate a hypothesis or were inconclusive, I went in with nothing but a brainstorming topic. But every time — every single time — I would go to meet with my advisor, by the end of that one-hour chat in a small office where the only tools we used to communicate were a pen and paper and a whiteboard, I would feel a rush of excitement, immediately get back to my desk, and get started on a brand new set of hypotheses. Every single time, I had spent a sleepless night trying to prepare for the meeting, and I would go back to the drawing board as soon as the meeting was over. That’s how stimulating the discussions were, and how large the scope of possibilities for where we could go next in terms of narrowing down a problem worth solving, and a solution worth building. My advisor had figured out what motivates me — novel tasks, uncharted territory, and using obscure math to reframe an applied problem statement into something I could produce practical results on. When I’d get concerned about progress, he would refocus the conversation back to the technical problem, and he knew it worked to reinvigorate inspiration every single time. All throughout the years.

Glad to be part of this. Gemini Safety team 🚀 #Gemini #LLMs

Bert, Bard, and Beyond 🤔

Google's BARD's image understanding is amazing!🤯 Vision-language models leverage understanding from both vision and language modalities. #AI #LLM Here's a video of me prompting BARD with only a product image and it describes the product's key features in detail. #GenerativeAI

น้องสาวถามว่า สายการบินนี้เค้าเชียร์ลุงหรอ คือพอคิดปุ้ปแม่งมองเป็นอย่างอื่นไม่ได้ละ 5555555555555 #เลือกตั้ง2566 #ลุงตู่

Glad to be a part of #PaLM2 🌴 and #GoogleBard 🤖 announced at #GoogleIO I am very lucky to be working on something huge and directly related to my phd research, both LM and RL 🚀 Stay tuned, more are coming! blog-google.cdn.ampproject.o…

Glad to be a part of this huge effort!! 🤩

this is kinda gate-keepy, @icmlconf

Never expect to reach 500 citations before. I know it's not a big deal for senior researchers but it means a lot to me as a moderate PhD student. I still remember the moment I got my first citation. Extremely grateful to those who support me so far @Georgetown & @MassiveData_GU

One of the future work presented in my PhD defense is to watch the World Cup Final. I am proceeding 😎⚽️🔥 #WorldCup #ArgentinaVsFrance

Pass PhD thesis defense: ✅ Submit the dissertation: ✅ Start working full time: ☑️

Pass PhD thesis defense: ✅ Submit the final draft: ☑️ Start working full time: ☑️

📣 Calling all @georgetown students! @MassiveData_GU is pleased to announce a call for applications for MDI Scholars Spring 2023 cohort! More info application can be found here ➡️ mdi.georgetown.edu/news/spri… Applications due by Friday, December 9th at 11:59pm Eastern.

Just submitted my PhD dissertation draft. Hope everything goes as planned 🤞 #phdlife

MOSAIC PROJECT: Covid-19 decreasing as a major financial threat. No surprise - inflation is largest future concern. mosaic.mdi.georgetown.edu/re…

Finally, I'm at the point that I read a related work section and noticed that it says the author names of some papers incorrectly... 🧐

Work on #Politics and #NLP? 🧐 Esp, Twitter data? We release language models pre-trained on >80 million US politics-related tweets. #LREC2022 Paper: aclanthology.org/2022.lrec-1… Repo: github.com/GU-DataLab/PoliBE…

Some ML research areas are extremely crowded. Working there is a race. But it's not about "how to benefit science" but "who gets the credit." This is the furthest from being a scientist I can imagine. How did we get here?

Attened 2 live talks from 2 of my (academic) idols on the same day. One from Prof @ylecun at @ECMLPKDD and the other one from Prof @KaiShu0327 at @Georgetown. What a day! 🤩 #ecmlpkdd2022 #AcademicTwitter