Olufunke Sarumi

Just Me|Love God|Positive Energy |NLP Scientist

[Natural Language Processing Papers @HEI]

Fine-Grained Perspectives: Modeling Explanations with Annotator-Specific Rationales

Olufunke O. Sarumi, Charles Welch, Daniel Braun
https://t.co/HL9UXwtgmj [𝚌𝚜.𝙲𝙻 𝚌𝚜.𝙰𝙸]
💬Accepted at 5th NLPerspectives Workshop

[Brigitte Gabriel @ACTBrigitte]

The Muslim Brotherhood's plan to destroy America has been in action for over 30 years and we are now witnessing the results.

Terror-linked groups like CAIR are celebrating the wins of 42 Muslim candidates in the 2025 election, and this is just the beginning.

Wake up, America!

[Christina Buttons @buttonslives]

I went to a church service for the first time last weekend, along with two vigils for Charlie Kirk. I was raised secular and I’m an atheist, but when he was killed, I felt a strong desire to be around people like I never had before.

At the service, the appeal became clear. What must it feel like to go through life never truly alone — to feel connected to someone who loves you and to believe you serve a higher purpose?

It hit me then just how alone I have been my entire life. From an early age, I was disconnected from everyone. I became depressed as a child, and that turned into severe depression in adolescence. I went into self-destruct mode and my life went off the rails. But even after I got my life on a better track in my twenties, and developed a relationship with my parents, I still kept people at a distance.

In church, seeing people so happy being connected to each other and to their God, it became apparent that this is likely a big reason conservatives tend to have better mental health than liberals. I can’t know how my life would have gone had I been raised differently, but I do wonder about it.

I’m skeptical that I’ll suddenly start believing in something I’ve never believed in. But people who grow up with values like family, faith, and community — and keep them — seem to be much better off. That’s something worth advocating for, and something I will support going forward.

[Olufunke Sarumi @funzac83]

@lucie_nlp lucie means amazing💖

[Cohere Labs @Cohere_Labs]

@singhshiviii @freddiev4 @mrdanieldsouza @tellarin @freakynut @weiyinko_ml @krypticmouse @DeividasMat @mjjzha @ramith__ @josephwilsonca @_marinafmachado @luisa_moura_ @dokatox @irombie @IfyOD1 @zaidalyafeai @chien_vu1692 @seb_ruder @surya03gsk @Emad_A_Alghamdi @sebgehr @Muennighoff @max_nlp @KreutzerJulia @mziizm @sarahookr @ahmetustun89 @viraataryabumi @yong_zhengxin @lekeonilude @NeelBhandari9 @amrmkayid @ShayneRedford 3️⃣ SCHOLARS PROGRAM - We’ve welcomed two cohorts of Scholars based across Germany, Brazil, India, Nigeria, Canada, and the United States. These emerging researchers now have 14 research papers to their names. 🔥 Congratulations!

[Irene Chen @irenetrampoline]

The transition to assistant professor made me feel happy, free, confused, and lonely at the same time. I wrote a blog post about my first 100 days as a professor including things I tried that did or didn't stick

[Olufunke Sarumi @funzac83]

🎉 Just had an incredible experience attending 2024 Annual Conference of the North American Chapter of the Association for Computational Linguistics! 🎉 - via #Whova event app

[Matthias Orlikowski @morlikow]

"Capturing Perspectives of Crowdsourced Annotators in Subjective Learning Tasks"

by @negar_mkhbrn and co-authors such as @Ginger_in_AI

Paper: https://t.co/4QRPXEUSx4

Post: https://t.co/kCiFyENa3F

[Matthias Orlikowski @morlikow]

"Corpus Considerations for Annotator Modeling and Scaling"

from @lucie_nlp's group @CaisaLab with first author @funzac83

Paper: https://t.co/eo5AUbOnL7

Post: https://t.co/Nw7cT1WDRw

[Lucie Flek @lucie_nlp]

In (2), @shaina_ashraf together with Fabio Gruschka and @ytaipsw show that detecting unreliable news spreading in Reddit and Twitter is more accurate when including behavioral patterns of the users

[Lucie Flek @lucie_nlp]

Check out 3 @CaisaLab papers @ #NAACL : 
(1) Scaling annotator modeling,
(2) modeling information spreader behavior,
(3) group fairness preservation under differential privacy.
Underline links:
1) https://t.co/myhBDLlEZY
2) https://t.co/Mcpcd1m5Mu
3) https://t.co/bP3OgwIzlc

[CAISA Lab @CaisaLab]

We're proud to have our #CAISA team presenting their work at #LREC2024 🌐✨ Join them in discovering their innovative contributions to #NLP and #AI.

[Olufunke Sarumi @funzac83]

Thanks@lucie_nlp

[Lucie Flek @lucie_nlp]

Our paper „Corpus Considerations for Annotator Modeling and Scaling“ accepted to #NAACL2024 . More info to come, congrats to the authors!

[AerIn @aerinykim]

I still remember being surprised 8 yrs ago when I first reduced Word2vec/Glove embeddings to a lower dimensional space. To my surprise, antonyms were placed very close to the words themselves and models sometimes failed to understand negation. At the time, I couldn't find a way to disentangle embeddings without distorting the entire representation. I moved on, thinking their proximity was from their overlapping context windows.

To my amazement again, this pattern has resurfaced in my daughter's early speech. She often amusingly confused words with their opposites, such as saying "TV is noisy!" when I muted it or "I don't want water. I'm already thirsty." when she meant she's hydrated.

Her adorable mix-ups validate to me that the inductive bias of Word2vec/glove's cost function was indeed correct, which is all about co-occurence and surrounding windows. It captured how humans learn language, which is a building block for AGI.

When GPT4 was released, negation was the first thing that I tested, and I've never had any issues with it since. Could scaling have solved the negation problem? Possible. Back then word2vec and glove was only 300 dim despite their strength. :)

[Olufunke Sarumi @funzac83]

@lucie_nlp 😅

[Olufunke Sarumi @funzac83]

Can you accept my invitation
https://t.co/HY93rb7kda

[Olufunke Sarumi @funzac83]

@lucie_nlp @CaisaLab interesting vibes🥰

[Lena Maier-Hein New: @lena-maier-hein.bsky.social @lena_maierhein]

Really looking forward to the talk of rock star @QiDou_ in our data science seminar next week!

Zoom link see: https://t.co/t46jni8qeL

[Rev. Olusola Areogun @ReverendAreogun]

Obedience is what Christianity is all about! Ask yourself, what current divine instructions are you obeying?

You will need to do a quick review of your instructions today. The most dangerous point you can be as a believer is when God is waiting for your obedience. That time, you are in no man's land.

#ReverendAreogun
#Obedience #Manifestation

[Olufunke Sarumi @funzac83]

@dolawanle God bless you for this post!