Prof. at Tohoku University, Sendai 🇯🇵. Natural Language Processing, Machine Learning, Psycho&Neurolinguistics. ex. @allen_ai @jhuclsp @NAIST_MAIN_EN
Joined May 2014
- Tweets464
- Following425
- Followers1.2K
- Likes838
I’m attending #IROS2026 in Pittsburgh! This will be my first presentation at a robotics conference. Looking forward to meeting people and learning more about the robotics community.
Keisuke Sakaguchi retweeted
A long-held dream, finally real: JSALT 2027 comes to Japan!!
Tohoku University, Sendai, 14 Jun – 6 Aug 2027. One problem, one team drawn from academia, industry and government, in residence. A first in Japan in three decades.
1-page proposals due 16 Oct: forms.gle/Q6ZSxkJtLLRZcnJQ9
Keisuke Sakaguchi retweeted
JSALTがついに日本に来ます！長年の夢がついにかないました。日本の研究者の皆様、ぜひ奮ってご応募ください！一緒にJSALTを日本で盛り上げましょう！
A long-held dream, finally real: JSALT 2027 comes to Japan!!
Tohoku University, Sendai, 14 Jun – 6 Aug 2027. One problem, one team drawn from academia, industry and government, in residence. A first in Japan in three decades.
1-page proposals due 16 Oct: forms.gle/Q6ZSxkJtLLRZcnJQ9
Keisuke Sakaguchi retweeted
Kotoba is teaming up with @KindredVentures for a night on the frontier of Voice AI for East Asia 🎙️We'll cover our datacenter + on-device speech AI, our speech-to-speech simultaneous interpretation model, and the Kotoba app — now with hundreds of thousands of users in East Asia — plus where voice agents go 4pm next Wed, June 24 · San Francisco
RSVP 👇
luma.com/bosc254j
#VoiceAI #GenAI #Speech #OnDevice
Keisuke Sakaguchi retweeted
We hope this enables the community to study native uniform diffusion at scale, and to dig into behaviors like these, among its many still poorly understood aspects. 🧵fin.
Huge thanks to my co-authors!
@k8kudo, @wataruuuuu_i, @VolumeisRyo, @KeisukeS_,
@drJunSuzuki
Keisuke Sakaguchi retweeted
Uniform diffusion LMs can revise any token freely, perhaps diffusion's most flexible paradigm. Yet no open model was trained from scratch at both large parameter and token scale.
We introduce Sumi, the first: 7B params, 1.5T tokens. Fully open. 🧵1/6
🔗 nlp.ecei.tohoku.ac.jp/projec…
Keisuke Sakaguchi retweeted
My new reflection: Language, Curiosity, and Life.
It is my attempt to put into words what has mattered most to me: language, family, curiosity, music, work, illness, and gratitude.
This is not a goodbye. I hope the conversation continues.
masatohagiwara.net/lcl.html
Keisuke Sakaguchi retweeted
Grateful to collaborate with @k8kudo, @y_aoneko, @rtanaka_lab, @saito_itsumi, @KeisukeS_, and @inuikentaro.
Thank you for the great support and insightful discussions!
Keisuke Sakaguchi retweeted
🔍 Key findings:
- Node and global information are linearly encoded within single image patches in the vision encoder.
- Edge information only becomes linearly encoded for the first time within text tokens in the language model.
2/N
Keisuke Sakaguchi retweeted
🚀 New paper on arXiv!
arxiv.org/abs/2603.02865v1
🤔 How do LVLMs internally form representations of nodes and edges?
💡 Node representations form early, whereas edge representations form late.
Feedback is welcome! 1/N
🚀🚀🚀
Welcome to Tokyo and Inspired.Lab @huggingface !
We’re looking forward to collaborating and having many great interactions!
Keisuke Sakaguchi retweeted
I’m currently attending #ICLR2025 and will be presenting our work today on sketch-to-diagram generation with vector code translation ✍️📐.
📍 See you at Poster Session 4, #82 (Friday 3-5:30pm)!
More details below 👇
Tired of the hassle of crafting figures and diagrams? Sketch2Diagram is our step toward turning hand-drawn sketches into structured, vector-based outputs—check it out if you're curious!
We’ll be at Poster #82 at 3PM — come say hi 👋 and check out our work! #ICLR2025
I’m currently attending #ICLR2025 and will be presenting our work today on sketch-to-diagram generation with vector code translation ✍️📐.
📍 See you at Poster Session 4, #82 (Friday 3-5:30pm)!
More details below 👇
Keisuke Sakaguchi retweeted
I'm thrilled to share that our Delphi paper is officially published today at @NatMachIntell after almost four years of hard works from all my amazing collaborators (a quite insane timeline considering the rapid AI world)! Special thanks to the unwavering support of my advisor, the marvelous @YejinChoinka! This project has defined my PhD experience and profoundly shaped how I view and approach science. The challenges along the way have made this journey all the more rewarding; it has been a truly unique and transformative training experience—one that anyone could aspire to during their PhD! nature.com/articles/s42256-0…
Keisuke Sakaguchi retweeted
Thrilled to see @JohnsHopkins affiliated with 2 of the 4 🎖️outstanding🎖️ paper awards at #COLM2024 🥳
Dated Data @jeff_cheng_77 @ruyimarone @orionweller @DanielKhashabi @ben_vandurme in @jhuclsp
Auxiliary Task Demands @_jennhu incoming prof in @JHUCogSci
Keisuke Sakaguchi retweeted
🏆🏆🏆🏆 Four papers at COLM were awarded outstanding paper awards. Congratulations to the authors!
Keisuke Sakaguchi retweeted
Kotoba has been accepted to the Japanese government's GenAI program! Will be building speech foundation models with GPU support.
Speech is exciting. I feel like I'm back in my first year as a PhD student. Check our application form if you're interested:
docs.google.com/forms/d/e/1F…
Kotoba Tech.は、音声基盤モデル開発を加速させる為に、GENIACに新たな事業者として採択されました。大規模GPU資源を活用して、音声という切り口から日本のAI開発に貢献して参ります。当社では一緒に東京オフィスで開発を進めていくメンバーを広く募集しています。Let's build faster!
meti.go.jp/policy/mono_info_…
Keisuke Sakaguchi retweeted
Kotoba Tech is launching the Open-GPT-4o project with a focus on speech foundation models. Let's make this happen fast with the power of open-source!
github.com/kotoba-tech/Open-…
We've developed two main building blocks: Kotoba-Speech (an autoregressive speech generation model) and Kotoba-Whisper (fast, accurate Japanese ASR).
As an intro to speech models, Jungo (@jungokasai) gives a quick tutorial on the audio tokenization technique, which allows us to train models similarly to text LLMs.
youtube.com/watch?v=2vu6u5Cr…
Exciting announcements coming soon from Kotoba!
Keisuke Sakaguchi retweeted
Kotoba-Speech v0.1 is here!
Transformer音声基盤モデル、Kotoba-Speech v0.1のデモ・モデルを公開しました！Kotoba Technologiesでは、音声・マルチモーダル基盤モデルを大規模に開発していきます。今後もご期待ください。
Demo: shorturl.at/cjmAC
Discord: shorturl.at/pzEL8
Model: shorturl.at/oyK16
Keisuke Sakaguchi retweeted
Kotoba-Speech v0.1: our first step towards speech, multimodal foundation model. Demo/HuggingFace Model are available. We'll be releasing our training code very soon!
@kotoba_tech #NLP2024
Transformer音声基盤モデル、Kotoba-Speech v0.1のデモ・モデルを公開しました！Kotoba Technologiesでは、音声・マルチモーダル基盤モデルを大規模に開発していきます。今後もご期待ください。
Demo: shorturl.at/cjmAC
Discord: shorturl.at/pzEL8
Model: shorturl.at/oyK16