Alexandre Défossez
@honualx.bsky.social
📤 141
📥 59
📝 9
Chief Exploration Officer
@kyutai-labs.bsky.social
in Paris.
We just released
unmute.sh
🔇🔊 It is a text LLM wrapper, based on in-house streaming ASR, TTS, semantic VAD to reduce latency. ⏱️ Unlike Moshi 🟢, Unmute 🔊 is turn base, but allows customization in two clicks🖱️: voice and prompt! Paper and open source coming soon.
loading . . .
6 months ago
1
9
1
We just open sourced a fine tuning codebase for Moshi!
add a skeleton here at some point
8 months ago
0
4
0
Just back from holidays, so a bit late, to announce MoshiVis, extending Moshi's multimodal capabilities to take in images 📷. Only 200M weights were added to plug a ViT through cross attention with gating 🖼️🔀🎤 Training relies on a mix of text only and text+audio synthetic data (~20k hours) 💽
add a skeleton here at some point
8 months ago
0
3
2
I'll start my presentation in 10 minutes, you can join in Zoom:
concordia-ca.zoom.us/j/81541793947
See you there!
add a skeleton here at some point
8 months ago
0
0
0
I'll present a dive into Moshi 🟢 and our translation model Hibiki 🇫🇷♻️🇬🇧 as part of the next
@convai-rg.bsky.social
reading group 👨🏫📗. 📅 13th of March 🕰️ 11am ET, 4pm in Paris. I'll discuss Mimi 🗜️ and multi-stream audio modeling 🔊. Join on Zoom, replay on YT. ⬛ ⬛ 🟧 🟧 🟨 🟨 🟩 🟩 🟩 ⬛ ⬛ 🟧 🟧 🟨 🟨 🟩 🟩 🟩 ⬛ ⬛
add a skeleton here at some point
9 months ago
0
6
2
reposted by
Alexandre Défossez
Kyutai
9 months ago
Even Kavinsky 🎧🪩 can't break Hibiki! Just like Moshi, Hibiki is robust to extreme background conditions 💥🔊.
loading . . .
0
8
5
reposted by
Alexandre Défossez
Jean-Rémi King
9 months ago
Very happy to have participated in this *beautiful* documentary from Florent Muller, on the frontiers between humans and machines, following next
@yann-lecun.bsky.social
and so many humbling figures of AI:
www.france.tv/documentaire...
loading . . .
France TV - Replay et Direct tv des chaînes France Télévisions (ex Pluzz)
Retrouvez toutes les vidéos, articles et podcasts des programmes des chaînes de France Télévisions.
https://www.france.tv/documentaires/documentaires-societe/l-homme-a-la-machine/
0
7
2
reposted by
Alexandre Défossez
Jean-Rémi King
9 months ago
Our latest studies on the decoding text from brain activity, reviewed by MIT Tech Review
@technologyreview.com
www.technologyreview.com/2025/02/07/1...
loading . . .
0
18
7
reposted by
Alexandre Défossez
Quentin Berthet
9 months ago
Check out our paper, with Lawrence Stewart and
@bachfrancis.bsky.social
Link:
arxiv.org/abs/2502.02996
1/8
loading . . .
Building Bridges between Regression, Clustering, and Classification
Regression, the task of predicting a continuous scalar target y based on some features x is one of the most fundamental tasks in machine learning and statistics. It has been observed and...
https://arxiv.org/abs/2502.02996v1
1
9
2
Excited to meet and exchange with a number of actors from all around the world at the AI Summit 🌍
9 months ago
0
2
0
We just released Hibiki, a 🎙️-to-🔊 simultaneous translation model 🇫🇷🇬🇧 We leverage a large synthetic corpus synthesized from the text translation model MADLAD, and our own TTS + simple lag rule. Model is decoder only, runs at scale, even on device 📲
github.com/kyutai-labs/hibiki
add a skeleton here at some point
10 months ago
0
1
0
reposted by
Alexandre Défossez
Jean-Rémi King
10 months ago
🚨Job alert (Please RT) What: masters internship and/or PhD positions Where: Rothschild Foundation Hospital (Paris, France) Topic: AI and Neuroscience Supervised by: Pierre Bourdillon and myself Apply here:
forms.gle/KKnea2QAjhAe...
Deadline: Feb 5th
loading . . .
https://forms.gle/KKnea2QAjhAesJ9h7
0
15
11
We just released the Helium-1 model , a 2B multi-lingual LLM which
@exgrv.bsky.social
and
@lmazare.bsky.social
have been crafting for us! Best model so far under 2.17B params on multi-lingual benchmarks 🇬🇧🇮🇹🇪🇸🇵🇹🇫🇷🇩🇪 On HF, under CC-BY licence:
huggingface.co/kyutai/heliu...
add a skeleton here at some point
10 months ago
0
25
8
you reached the end!!
feeds!
log in