Rishu Kumar (@rishdotuk) header
Rishu Kumar (@rishdotuk) profile picture

Long document understanding, Multilingual Evals and efficient models mainly, but other #NLProc applications in free time | vim enthusiast

Joined April 2011
Photo by Rishu Kumar (@rishdotuk)Photo by Rishu Kumar (@rishdotuk)Photo by Rishu Kumar (@rishdotuk)Photo by Rishu Kumar (@rishdotuk)Photo by Rishu Kumar (@rishdotuk)Photo by Rishu Kumar (@rishdotuk)Photo by Rishu Kumar (@rishdotuk)Photo by Rishu Kumar (@rishdotuk)Photo by Rishu Kumar (@rishdotuk)Photo by Rishu Kumar (@rishdotuk)
Rishu Kumar retweeted
I have been handed an object of immense power
129
745
39
11,664
149,740
Rishu Kumar retweeted
was bitten by a radioactive lawyer. woke up with power of attorney.
40
1,541
45
14,376
148,183
Rishu Kumar retweeted
Never put all your eggs in one basket
525
4,548
539
64,259
1,313,024
Rishu Kumar retweeted
Baniyabaazi is holding India back. Everyone is busy making a chutiya out of everyone else, little realizing that in the process, they’re making a chutiya out of themselves. When everyone tries to outsmart everyone else, the entire nation loses.
223
805
55
5,234
104,241
JUST IN: Riemann Hypothesis has been solved. alpo.ge/remark.pdf
1,820
8,108
5,369
61,185
15,281,122
Rishu Kumar retweeted
on-device FTW
Dictation in ChatGPT is getting faster and more reliable. With Live Dictation, your speech turns into text as you talk, so you spend less time waiting for your message to finish transcribing. Yap away!
3
5
37
1,408
Rishu Kumar retweeted
Just prompting Claude can tell you if a text is written by AI, if yes which AI, and if no, often even which human wrote it! no web access, tools, or candidate list required... Authorship attribution is advancing extremely quickly, with major deanonymization/impersonation risk.
Frontier models need only one text to unmask its author. Give a frontier model a single text, and it can often tell whether AI wrote it, which lab's AI, or even the specific person behind it. No web search, no list of candidates, with one text. 🧵
3
10
62
6,374
Rishu Kumar retweeted
I want to thanks @SemiAnalysis_ for the motivation. In the Gemini RL scaling team, we printed & put that article on our desks to be reminded daily of what used to be true at the time. But the times they are changing :)
110
412
82
6,979
580,188
Rishu Kumar retweeted
Esto es precioso. Un chico en redes sociales compartió que su mejor amigo falleció hace 4 meses. De vez en cuando se cruza con su Driveatar en Forza Horizon y conducen juntos.
44
622
28
7,495
305,680
Some words of kindness to combat all the online vitriol 🙃
2
1
60
2,827
no change November 😔#NOCHANGENOVEMBER Do not learn 🚫! NO PERSONAL GROWTH ❌ ! NO CHANGE!!!!
17
1,029
22
8,600
137,718
Rishu Kumar retweeted
The hydrated will inherit the earth
13
51
8
771
46,208
Rishu Kumar retweeted
AI is killing the old web. When an AI overview is available, Google searches result in a web site visit 40% less often. Obvs this is better for the user, but the old view-driven model of the web is in trouble.
4
11
7
86
14,004
Rishu Kumar retweeted
भोजन को भोजन से लड़ाने वाले। भाषा को भाषा से भिड़ाने वाले। जेंडर को जेंडर से लड़ाने वाले। इतिहास को वर्तमान के सामने खड़ा करके दोनों के हाथ में पत्थर थमाने वाले। आस्थाओं को आस्थाओं के ख़िलाफ़ भड़काने वाले। जातियों को जातियों की चौकीदारी में लगाने वाले। स्त्रियों को स्त्रियों की स्वतंत्रता से डराने वाले। पुरुषों को पुरुष होने के अहंकार के बीच झुलाने वाले। पहनावे को विचारधारा का प्रमाणपत्र बनाने वाले। रसोई को रणभूमि में बदलने वाले। त्योहारों को शक्ति प्रदर्शन में बदलने और खानपान में षड्यंत्र सूँघने वाले। कलाकार को कला से अलग करने वाले। विश्वविद्यालयों को विचारधाराओं की छावनी बनाने वाले। विज्ञान को आस्था की अदालत में खड़ा करके माफ़ीनामा लिखवाने वाले। बहुमत को महानता और अल्पमत को अपराधबोध में बदलने वाले। हर असुविधाजनक सवाल के सामने एक सुविधाजनक दुश्मन खड़ा करने और प्रेम को प्रेम के सामने संदिग्ध बनाने और पूरी उम्र नफ़रत के लिए लोगों को जोड़ते रहने वाले बताएँगे कि प्रेम कैसे होगा?
8
23
93
3,470
Rishu Kumar retweeted
Important public service announcement for writers, from @brunellaism
2
21
160
5,756
Saw this on LI. Hilarious! 😆😆
5
32
1
273
7,079
It’s really depressing to see so much hate directed at Tuhin for the crime of daring to say true things that authoritarians dislike.
It’s unfortunate that mostly black authors have been outed for using AI to write fiction, but does that mean AI detectors are racist ? Sadly, the answer is a resounding NO. With the help of @pangram and @GPTZeroAI I analyzed 187 African books published between 2001 & 2026. 👇👇
1
4
110
6,033
Rishu Kumar retweeted
Agree we need more data, it's hard to get, and when someone tries fair evaluation (@TuhinChakr), they get all the hate. I'm not sure which one it is, because Tuhin is criticized for doing evaluation on target variety and domain, while Pangram is criticized for not doing it...
I saw someone trying to claim that pangram is basically infallible because OMG they had FOUR datasets of NON NATIVE ENGLISH. Like "non native English" can be flattened and captured by these datasets. Y'all are truly incredible it's 100% the twilight zone.
5
4
79
3,804
It’s unfortunate that mostly black authors have been outed for using AI to write fiction, but does that mean AI detectors are racist ? Sadly, the answer is a resounding NO. With the help of @pangram and @GPTZeroAI I analyzed 187 African books published between 2001 & 2026. 👇👇
42
46
111
653
161,184
Rishu Kumar retweeted
Feels like one thousand years ago...
2
2
16
1,149