Sing Chen Yeo

Data Scientist and cognitive scientist - Developing AI with emotion - Opsis Emotion AI

[Andrej Karpathy @karpathy]

It's a bit sad and confusing that LLMs ("Large Language Models") have little to do with language; It's just historical. They are highly general purpose technology for statistical modeling of token streams. A better name would be Autoregressive Transformers or something.

They don't care if the tokens happen to represent little text chunks. It could just as well be little image patches, audio chunks, action choices, molecules, or whatever. If you can reduce your problem to that of modeling token streams (for any arbitrary vocabulary of some set of discrete tokens), you can "throw an LLM at it".

Actually, as the LLM stack becomes more and more mature, we may see a convergence of a large number of problems into this modeling paradigm. That is, the problem is fixed at that of "next token prediction" with an LLM, it's just the usage/meaning of the tokens that changes per domain.

If that is the case, it's also possible that deep learning frameworks (e.g. PyTorch and friends) are way too general for what most problems want to look like over time. What's up with thousands of ops and layers that you can reconfigure arbitrarily if 80% of problems just want to use an LLM?

I don't think this is true but I think it's half true.

[Kyle Kabasares @kylekabasares]

That feeling when ChatGPT o1 accomplishes in 1 hour what took you about a year in your PhD: https://t.co/jG7UxEUT12

[Lewis Tunstall @_lewtun]

Excited to release Zephyr-7b-beta 🪁 !

It pushes our recipe to new heights & tops 10x larger models 💪

📝 Technical report: https://t.co/3R4czrpbu5
🤗Model: https://t.co/8uUkvg4E7j
⚔️Evaluate it against 10+ LLMs in the @lmsysorg arena: https://t.co/2cMZRUvhOc

Details in the 🧵

[Sing Chen Yeo @yeo_sing]

How do students' sleep timing clash with #university scheduling policy? Check out my latest work on the effect of class start time and #chronotype on students' academic performance, #sleep, and class attendance
https://t.co/yLxagO0VRC

[Sing Chen Yeo @yeo_sing]

Time to graduate! Welcome anyone who is interested in #sleep and #learninganalytics research to join my public defense.

Please note that the date and time are in Singapore local time.

https://t.co/VK6SSiGJBC

[Sing Chen Yeo @yeo_sing]

It's cool to see that #estonia #news outlet is reporting my research on early classes and #sleep

Varahommikune loeng kärbib tudengi õpiedu

https://t.co/3UlvITslxp #ERR

#university #research #bigdata #science

[Sing Chen Yeo @yeo_sing]

Cool to see someone in Russia likes my research :)

Пары в восемь утра плохо сказались на успеваемости студентов: https://t.co/X7M21VajYx

[Sing Chen Yeo @yeo_sing]

Nice to see my research on @newscientist. University students with morning lectures tend to have lower grades https://t.co/mBQes3TYeq

[Sing Chen Yeo @yeo_sing]

Proud to share the first chapter of my thesis: using big data to show that early morning classes are bad for your attendance, sleep, and academic performance.

#University #learning #DataScience
https://t.co/XdetQ9OCCh

[NUS ALSET @alsetNUS]

A small number of #eLearning classes can decrease the potential for disease transmission while minimising disruption to university operations 🎓 A great paper by @yeo_sing et al. using the @alsetNUS Data Lake https://t.co/ILgUnGHOjP @OER_NIE @CherelEric @Pier_Grd @SoLAResearch

[Sing Chen Yeo @yeo_sing]

My latest publication on evaluating #university safe-measurement for #COVIDー19 is available: https://t.co/0cDCCnGN05
#research #BigData

[Sing Chen Yeo @yeo_sing]

My preprint is featured on the @ResearchGate #COVID19 community page. Read it here: https://t.co/w8JzItf3fM

[Sing Chen Yeo @yeo_sing]

My preprint is featured on the @ResearchGate #COVID19 community page. Read it here: https://t.co/w8JzItf3fM

[SoLAR @SoLAResearch]

We are so pleased to be able to support 31 students to attend #LASI19! Thank you to our #LAK19 sponsors, @MHEducation, @CengageLearning, @CanvasLMS, @ACT and @Blackboard who play a pivotal role in providing this funding to expand the field of #LearningAnalytics! #ErikDuval

[Neil deGrasse Tyson @neiltyson]

Arc of Enlightenment in a rational, civilized world:

Data → Facts → Information → Knowledge → Understanding → Wisdom.

[Neil deGrasse Tyson @neiltyson]

SCIENTISTS: “We’ve produced the first-ever image of a supermassive Black Hole, 55-million light years away”
RESPONSE: “Oooh!”

SCIENTISTS: “We’ve concluded that humans are catastrophically warming Earth”
RESPONSE: “That conflicts with what I want to be true, so it must be false”