
Japan, AI, NLP | Professor @KitamiOf | Senior member @IEEEorg | Automatic #Cyberbullying Detection | #Ainu Language Processing | #Coffee Science
Japonia
Joined June 2019
- Tweets2.6K
- Following328
- Followers256
- Likes4.2K
🚨 The first scientifically proven #language #similarity metric - qWALS - has been published. Please, try it out in your research.
#langaugesimilarity #linguisticdistance #qwals #wals #WorldAtlasOfLanguageStructures #linguistics #NLProc #NLP
pypi.org/project/qwals/
🚨 How much exactly are two languages similar?
…and how can language models use that information?
We propose first actually usable language similarity score: qWALS. Language models can use it to get better results even with ZERO data in target language❗️
t.ly/9nS_d
Michal Ptaszynski | プタシンスキ ミハウ retweeted
On 30 Nov, I’m speaking at MOST 2026 “AI without Borders” at IPI PAN, a session bringing together Polish AI researchers working abroad. Hybrid, in English, free registration; you can join online for a single talk: most-pan-2026.github.io/
But… can it play Doom? 🤔
Meet qwenGPU: Qwen3.5-2B with a SIMT GPU inside its weights, 45 compiled layers in front of Qwen’s own, using its Gated DeltaNet layers as registers and memory.
It runs real CUDA in its forward passes. The host compiles the kernel to PTX and loads it, then the model writes every step of the launch itself: every register, branch and memory write.
The GPU has 4-lane warps, a 4-stage pipeline, 15 registers per lane and 126 words of global memory, all held in DeltaNet state. Qwen generates each launch token by token, each record running one machine block for one warp.
34
Michal Ptaszynski | プタシンスキ ミハウ retweeted
Call for papers: IntelliSym2027, an AAAI-27 workshop on intelligence symbiosis. How do norms and trust infrastructure emerge among AI agents and humans? Deadline Nov 20 (fast track for AAAI-27 rejects: Dec 3). Montréal, Feb 22/23. intelligence-symbiosis.net/e… #AAAI2027
Michal Ptaszynski | プタシンスキ ミハウ retweeted
🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡🤡
telegraph.co.uk/world-news/2…
Gratulacje. 👏 Chętnie zobaczyłbym coś więcej niż filmik promocyjny, np. otwarty raport, albo pracę naukową na podstawie badań przeprowadzonych z pomocą tego narzędzia.
Razem z @sentionepl prezentujemy Narrative IQ, pierwsze na świecie narzędzie do przeciwdziałania systemowej dezinformacji.
Działa 24/7 analizuje miliony wzmianek.
Identyfikuje źródła i kanały, którymi rozchodzi się dezinformacja, i kieruje automatycznie rzetelną odpowiedź tam, gdzie jej zasięg rośnie najszybciej, ograniczając dalszą dystrybucję algorytmiczną.
To nowy standard ochrony organizacji.
Od grudnia w dostępie komercyjnym.
23
Michal Ptaszynski | プタシンスキ ミハウ retweeted
$52M seed round. 15M+ users. But this is what we're most proud of yet.
Introducing Drama 3: the most emotive and controllable TTS model in the world
And yes, it's open-weight.
You can now control how even a single word should sound, change its tone, and shift your voice mid-sentence.
Uncle Roger was impressed too.
Try it here:
s.fish.audio/rqh0l8
Comment “DRAMA” within the next 48 hours for 1 month free.
If you're an enterprise your first month is on us:
s.fish.audio/qz9r0t
Michal Ptaszynski | プタシンスキ ミハウ retweeted
The Claude Code hackathon is back, and we're bringing it to Japan!
エンジニアの方も、そうでない方も、日本在住のビルダーならどなたでも応募できます。世界に何を見せてくれるのか、楽しみにしています。
応募締切は10月26日 23:59(日本時間):
cerebralvalley.ai/e/built-wi…
#claudejapan2026
All those lists have always one thing in common. They never explain what harassment actually is. They are, as one would say, ignotum per ignotum.
A Japanese law firm has prepared an instructive list of forms of harassment. The nature of some should be obvious, others maybe not.
Sexual Harassment (seku-hara)
Power Harassment (pawa-hara)
Maternity Harassment (mata-hara)
Paternity Harassment (pata-hara)
Alcohol Harassment (aru-hara)
Smoke Harassment (sumo-hara)
Moral Harassment (mora-hara)
Karaoke Harassment (kara-hara)
Smell Harassment (sume-hara)
Remote Harassment (rimo-hara) - harassment in connection with people who work from home
Technology Harassment (teku-hara)- harassment directed at people for not being good with technology
Restructuring Harassment (risutora-hara) - harassment directed at pressuring workers to quit.
Customer Harassment (kasu-hara) - harassment by customers
Enjoyment Harassment (en-hara) - harassment aimed at enjoying someone's enjoyment of something
Jitan-hara - telling people to work fewer hours while not doing anything to actually reduce the need to work and expecting the workers to figure it out
Owara-hara - pressuring candidates you want to work for you to stop looking for jobs elsewhere
Age Harassment (ei-hara) - harassment of older workers
Care Harassment (kea-hara) - harassment of workers for having to look after elder relatives
Academic Harassment (aka-hara) - harassment taking place in academia
Matching Harassment (machi-hara) - harassment conducted through matching apps.
Vaccine Harassment (waku-hara) - pressuring others to get various vaccinations
Jigo-hara - the harassment directed at someone who claimed harassment after th harassment investigation has concluded (i.e., retaliation?)
Logic Harassment (roji-hara) - use of logic as a tool of harassment.
Seems reasonably complete but is missing "gyaku-hara" (reverse harassment) which is when junior employees accuse your of power harassment.
komon-lawyer.jp/qa/harasumen…
55
Michal Ptaszynski | プタシンスキ ミハウ retweeted
Media reports claim that President Trump tried to persuade Poland's defence minister to leave the government and form a coalition with the right-wing opposition.
We outline what is known so far and how the government, including PM @donaldtusk, has reacted notesfrompoland.com/2026/10/…
Michal Ptaszynski | プタシンスキ ミハウ retweeted
Anyone travelling to #EMNLP2026 ? Please take a quick look and see if you could help!
We are looking at how people use machine translation tools in real-life situations when abroad.
Helping us would require you to participate in a 30-40 min interview (online) after the conference.
(if you can't help, can you share?)
Going to EMNLP 2026 in Budapest?
We are looking for 20-30 EMNLP attendees to help us understand how you handle language barriers during a conference trip.
Contact Yujun Wang: y.wang2.25@abdn.ac.uk
Please share!
@EhudReiter @mar_kar_ Pinzhen Chen, Yujun Wang
Fair point. But it shows Gebru and Bender couldn't imagine how fast LLMs would evolve in 3-4 years. And to be fair, most people in AI didn't either. The progress is indeed so impressive that we can't deny LLMs are closing in to be a pretty good approximation of human reasoning.
Calling LLMs "stochastic parrots" in 2021, then claiming Fable and Astra aren’t LLMs, is like arguing in 1900 that planes would never fly, then when they carry millions people, arguing that: "they aren’t really planes. look at all the tech around them - autopilot, radar, GPS"
59
I think by now we can all agree that every new paper on arxiv that is not a preprint from an actual publication is just AI slop. I can’t see how anyone would put anything of value there.
JEV was released in "limited early access" on September 15, that's two weeks ago. How come we have 29 papers about it on arxiv (and, I assume, ICLR submissions) already. This is not healthy.
1
87
Michal Ptaszynski | プタシンスキ ミハウ retweeted
9/26〜27は,北海道NLPミーティングをやっています.興味がある方は,ぜひYouTube Liveで,どうぞご参加ください.
プログラムは,HPに載せています.
Webpage: sites.google.com/view/hokkai…
YouTube: youtube.com/live/SMbmAoN2C6A…
As my super talented student @TyBrassington noticed the slop has no limits or no wonder this human has issues reading upside down checklist
Come on we can do better !
1
51
Finally! Keeping fingers crossed for you guys. 🤞
Tokyo, your autonomous ride is arriving in 2027 🇯🇵✨ Waymo, @Nihonkotsu_Taxi and @go_goesnext have signed an agreement preparing for the commercial launch of a fully autonomous ride-hailing service in Tokyo in 2027.
Read more: waymo.com/blog/2026/09/openi…
62





























