Kaustubh Olpadkar

ML Engineer at FinTech | Former Intern @BellLabs | Alumni @GradSBU @Themsubaroda. Views My Own!

[Kaustubh Olpadkar @olpadkar]

@karpathy This is super cool! Is there a similar tool for reading research papers or this works for papers as well?

[Kaustubh Olpadkar @olpadkar]

Huge thanks to my co-authors, mentors, and collaborators who made this work possible. I’m incredibly proud of what we’ve built and grateful to share it with the NLP community.

[Kaustubh Olpadkar @olpadkar]

I’m thrilled to share that our paper — Can LLMs Be Efficient Predictors of Conversational Derailment? — has been officially published in EMNLP 2025 Findings.

Read it here: https://t.co/26TyJpAxdb

[Tech At Bloomberg @TechAtBloomberg]

Our BLAW #AI Engineering team’s Kaustubh Olpadkar is presenting "Can LLMs Be Efficient Predictors of Conversational Derailment?" during #EMNLP2025's Virtual Poster Session today (8-9 AM CST)
https://t.co/un3cRys5ag
#NLProc #LLMs #GenAI
(1/3)

[Kaustubh Olpadkar @olpadkar]

We explore whether LLMs can forecast conversational derailment—spotting when a dialogue might go off-topic or toxic before it happens. Even without fine-tuning, LLMs act as lightweight, proactive moderators balancing efficiency and accuracy.

[Tech At Bloomberg @TechAtBloomberg]

Our BLAW #AI Engineering team’s Kaustubh Olpadkar is presenting "Can LLMs Be Efficient Predictors of Conversational Derailment?" during #EMNLP2025's Virtual Poster Session today (8-9 AM CST)
https://t.co/un3cRys5ag
#NLProc #LLMs #GenAI
(1/3)

[Kaustubh Olpadkar @olpadkar]

@Cloudwatch199 What about H1b renewals?

[Kaustubh Olpadkar @olpadkar]

@garykubiak @OpenAI Didn’t expect this 🥲

[Kaustubh Olpadkar @olpadkar]

@eaclmeeting Could you please remind the reviewers to take part in the discussion period? It’s one day remaining and I have not heard back anything on my rebuttal.

[Kaustubh Olpadkar @olpadkar]

Are the reviews for October 2023 cycles released yet? @ReviewAcl

[Kaustubh Olpadkar @olpadkar]

@eaclmeeting Decisions for #emnlp2023 are out in the first week of October. Can we upload the paper on arxiv after that and before submission deadline?

[Kaustubh Olpadkar @olpadkar]

@gimyong30482200 2 reviewers have not yet commented anything. One commented the same as yours with literally no change in review or scores. I don’t understand if it was changed or anything.

[Kaustubh Olpadkar @olpadkar]

Waiting for reviewers’ response. #EMNLP2023

[Kaustubh Olpadkar @olpadkar]

@mrdrozdov Reviewer 1: Reason to Reject: The motivation for the work are week.
Reviewer 2: Reason to Accept: The paper gives straightforward motivation for work.
😑 #emnlp2023

[Kaustubh Olpadkar @olpadkar]

@TuhinChakr For me, the reviewer is insistent on testing on a particular dataset and citing papers related to that which is on a very special domain even though I have proven my hypothesis on widely used nlp benchmarks. This is not the way to promote your datasets.

[Kaustubh Olpadkar @olpadkar]

@prajdabre1 @TuhinChakr I don’t see any option to flag the review on open review.

[Kaustubh Olpadkar @olpadkar]

@sonqt01 I have also got similar very frustrating reviews. Do you see option to reply or rebuttal ?

[Kaustubh Olpadkar @olpadkar]

@vageeshsaxena @emnlpmeeting Me too, so you see option for rebuttal?

[Kaustubh Olpadkar @olpadkar]

@emnlpmeeting are the reviews released? #EMNLP23

[Kaustubh Olpadkar @olpadkar]

@gerasimoss are reviews released? I have been checking since hours!

[Kaustubh Olpadkar @olpadkar]

@PrakamyaMishra @emnlpmeeting The email I got for emergency review clearly stated "Assignments have been made and emergency reviews are due August 11 EOD anywhere on earth."