Rishu Kumar

I love MT. :D

Must they torture me like this? :(

WAITER!!! WAITER!!!!
MORE CHAIN OF THOUGHT SURVEYS PLEASE!!!!

NOOOOOOOOOOOOO!!!

Screenshot of interaction with Gemini, converting my image to a random image.

Yes, ofcourse that's totally me. 😭😭 Am I using something wrong?

Narrator: He definitely let that slide. :D

Petition to change "Human evaluation" to "vibe evaluation" in summarization.

Off By One xkcd.com/3062

Each time I tried vibe-coding something for my nlp-related side quests, my respect for the companies and people shilling "Coding LLMs" dropped by 20%.

Rishu Kumar ‪Charles University, Faculty of Mathematics and Physics, Institute of Formal and Applied Linguistics‬ - ‪‪Cited by 2,157‬‬ - ‪Natural Language Processing‬ - ‪Linguistics‬

I will believe we have AGI when my Google Scholar profile gets fixed automatically and I am not shown as an Agricultural science researcher who has occasionally worked on NLP stuff.😭
scholar.google.co.in/citations?us...

The Factory must grow

Few obvious questions: Is the upper limit of ~30% hallucination acceptable? And how much work it adds to a lawyer’s work to ensure there is no hallucinations at all in the generated document?

(Haven’t read the entire paper yet)

Honest answer: I don't read anything from the enormous stream of papers unless they are *critical* to my current work or I am deeply interested in the topic.

Come present your #NLP work in Leuven!

Okay, automated cross-posting has its quirks. This was the original tweet: x.com/OpenAI/statu...

Reminds me of this from @xkcd


Though the real question is: how long before everyone races to optimize for this "benchmark?"

WHY ARE THERE 5 DIFF ELECTRON VERSION ON MY PC? :/

Sometimes VSCodium feels so slow that I just want to go back to jupyterlab and vim.

Next year we will reach NeurIPS level at this rate.

Twitter screenshot of https://x.com/ChombaBupe/status/1891070696832278614 which reads: Language models are still failing on trivial simple tasks, like following basic instructions, summarization, etc. Another screenshot of ars technical saying 'Over half of LLM-written news summaries have "significant issues" - BBC Analysis'

Honorary mention of arxiv.org/abs/2309.09558

If we start calling them anything else, it doesn’t anthropomorphise the models enough. :)

I still think about my application for an internship in 2022 where I got through the first screening, forgot about it, and got another email 6 months later in 2023, went ahead & did two(?) interviews, absolutely nailed it, got an email 6 months later that position is canned. :D

Amazon product name for Oral B Genius X electric toothbrush which has AI in it.

Does your brush even have AI, bro?

WHYYYYYYYYYY??

There must be a term for when someone takes your code and modifies the logic so unnecessarily and badly that even the second double espresso isn't enough to fix your mood.

I don’t think any of the “foundation model” orgs will ever talk about contamination, because that makes scores look bad. At this point a lot of these models seem like clowns in circus, only there to keep the public (investors?) at large excited about the circus that is “AI research”.

This might be the reminder I needed to fix my webpage. Side-quest for the day acquired. :p

ACL 2025 Ling theory & Cognitive modeling track emergency reviewer volunteer form The Linguistic Theories, Cognitive Modeling, and Psycholinguistics track at ACL 2025 is looking for emergency reviewers. The emergency reviews will take place between 18th to 26th of March, 2025. Thes...

Repost appreciated! 🙏

ACL 2025 Ling theory & Cognitive modeling track is looking for emergency reviewers. The emergency review period is between 3/18-26, and these reviewers will be excluded from the ARR cycle. If you're interested, please sign up here! docs.google.com/forms/d/1fH7...

I had this experience in my thesis and a research project that I did. Maybe I should publish that project on arxiv.

Watching a small model train on a consumer-grade GPU is where the fun at.

Didn’t realise it’s an imposter account. :D