Return-Path: Delivered-To: cadet@irif.fr Received: from mailhub.math.univ-paris-diderot.fr ([81.194.30.253]) by mailhost.irif.fr (Dovecot) with LMTP id eSn9JFkQfGHOEQAAP9ZUWg for ; Fri, 29 Oct 2021 17:16:41 +0200 Received: from mailhub.math.univ-paris-diderot.fr (localhost [127.0.0.1]) by mailhub.math.univ-paris-diderot.fr (Postfix) with ESMTP id 917DDD02F4; Fri, 29 Oct 2021 17:16:41 +0200 (CEST) X-Virus-Scanned: amavisd-new at math.univ-paris-diderot.fr X-Spam-Flag: NO X-Spam-Score: -1.629 X-Spam-Level: X-Spam-Status: No, score=-1.629 tagged_above=-10000 required=5 tests=[BAYES_00=-1.9, HEADER_FROM_DIFFERENT_DOMAINS=0.249, HTML_MESSAGE=0.001, MAILING_LIST_MULTI=-1, MISSING_HEADERS=1.021, RCVD_IN_MSPIKE_H2=-0.001, URIBL_BLOCKED=0.001] autolearn=no autolearn_force=no Received: from mailhub.math.univ-paris-diderot.fr ([127.0.0.1]) by mailhub.math.univ-paris-diderot.fr (mailhub.math.univ-paris-diderot.fr [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id lAdfUpE7rXWT; Fri, 29 Oct 2021 17:16:39 +0200 (CEST) Received: from korolev.univ-paris7.fr (korolev.univ-paris7.fr [194.254.61.138]) by mailhub.math.univ-paris-diderot.fr (Postfix) with ESMTPS id 72BC6D02ED; Fri, 29 Oct 2021 17:16:39 +0200 (CEST) Received: from mars.math-info.univ-paris5.fr (helios2.math-info.univ-paris5.fr [193.48.200.16]) by korolev.univ-paris7.fr (8.14.4/8.14.4/relay1/82085) with ESMTP id 19TFGdGF032553 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=OK); Fri, 29 Oct 2021 17:16:39 +0200 Received: from mars.math-info.univ-paris5.fr (localhost [127.0.0.1]) by mars.math-info.univ-paris5.fr (8.14.4/jtpda-5.4) with ESMTP id 19TFGb6P031793 ; Fri, 29 Oct 2021 17:16:37 +0200 Received: (from sympa@localhost) by mars.math-info.univ-paris5.fr (8.14.4/8.14.3/Submit) id 19TFGbkh031774; Fri, 29 Oct 2021 17:16:37 +0200 X-Authentication-Warning: mars.math-info.univ-paris5.fr: sympa set sender to diip-perimeter-owner@math-info.univ-paris5.fr using -f Received: from smtp-out02.parisdescartes.fr (smtp-out02.parisdescartes.fr [193.51.86.78]) by mars.math-info.univ-paris5.fr (8.14.4/jtpda-5.4) with ESMTP id 19TFGTCu031649 ; Fri, 29 Oct 2021 17:16:29 +0200 Received: from localhost (saroumane.univ-paris5.fr [192.168.253.9]) by mx1.parisdescartes.fr (Postfix) with ESMTP id D95782428ED; Fri, 29 Oct 2021 17:16:29 +0200 (CEST) X-Virus-Scanned: amavisd-new at univ-paris5.fr Received: from mataram.parisdescartes.fr ([192.168.253.4]) by localhost (fourmilier.univ-paris5.fr [192.168.253.9]) (amavisd-new, port 10024) with ESMTP id Q4WlMAgbNMih; Fri, 29 Oct 2021 17:16:27 +0200 (CEST) Received: from [192.168.0.13] (85-168-38-27.rev.numericable.fr [85.168.38.27]) by mataram.univ-paris5.fr (Postfix) with ESMTPSA id 8AF5F2FA01D; Fri, 29 Oct 2021 17:16:27 +0200 (CEST) Content-Type: multipart/alternative; boundary="------------eie7gDzqhJJivPBUNcS0Qn8J" Message-ID: <1b26f450-4798-a4dd-92e1-fba873a58dff@mi.parisdescartes.fr> Date: Fri, 29 Oct 2021 17:16:27 +0200 MIME-Version: 1.0 User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:91.0) Gecko/20100101 Thunderbird/91.2.1 From: Themis Palpanas Reply-To: "diip@math-info.univ-paris5.fr" Cc: "diip@math-info.univ-paris5.fr" Content-Language: en-US Subject: [diip-perimeter] rappel: diiP Distinguished Lecture, November 3, 4pm X-Loop: diip-perimeter@math-info.univ-paris5.fr X-Sequence: 32 Errors-to: diip-perimeter-owner@math-info.univ-paris5.fr Precedence: list Precedence: bulk Sender: diip-perimeter-request@math-info.univ-paris5.fr X-no-archive: yes List-Id: List-Archive: List-Help: List-Owner: List-Post: List-Subscribe: List-Unsubscribe: X-Greylist: Sender succeeded STARTTLS authentication, not delayed by milter-greylist-4.2.7 (korolev.univ-paris7.fr [194.254.61.138]); Fri, 29 Oct 2021 17:16:39 +0200 (CEST) X-Miltered: at korolev with ID 617C1057.000 by Joe's j-chkmail (http : // j-chkmail dot ensmp dot fr)! X-j-chkmail-Enveloppe: 617C1057.000 from helios2.math-info.univ-paris5.fr/helios2.math-info.univ-paris5.fr/null/mars.math-info.univ-paris5.fr/ X-j-chkmail-Score: MSGID : 617C1057.000 on korolev.univ-paris7.fr : j-chkmail score : . : R=. U=. O=. B=0.000 -> S=0.000 X-j-chkmail-Status: Ham This is a multi-part message in MIME format. --------------eie7gDzqhJJivPBUNcS0Qn8J Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit *diiP* * Distinguished Lecture:* *Promises and challenges of massive-scale AI – the case of large language models* *Who*: Laurent Daudet, CTO and co-founder at LightOn, Professor (on leave) of physics (Université de Paris) *When*: November 3, 4pm (Paris time). *Where*: hybrid: online (zoom) and at Room Turing Conseil, 45 rue des Saints Pères 75006 Paris zoom link: *title*: Promises and challenges of massive-scale AI – the case of large language models *abstract*: OpenAi’s GPT-3 language model has triggered a new generation of Machine Learning models. Leveraging Transformers architectures at billion-size parameters trained on massive unlabeled datasets, these language models achieve new capabilities such as text generation, question answering, or even zero-shot learning - tasks the model has not been explicitly trained for. However, training these models represent massive computing tasks, now done on dedicated supercomputers. Scaling up these models will require new hardware and optimized training algorithms. At LightOn - a spinoff of university research -, we develop a set of technologies to address these challenges. The Optical Processing Unit (OPU) technology makes some matrix-vector multiplications in a massively parallel fashion, at record-low power consumption. Now accessible on-premises or through the cloud, the OPU technology has been used by engineers and researchers worldwide in a variety of applications, for Machine Learning and scientific computing. We also train in an efficient manner large language models, such as PAGnol (demo at https://pagnol.lighton.ai  ), the largest language model in French, that can be used for various research and business applications. *short bio*: Laurent Daudet is currently employed as CTO at LightOn, a startup he co-founded in 2016, where he manages cross-disciplinary R&D projects, involving machine learning, optics, signal processing, electronics, and software engineering. Laurent is a recognized expert in signal processing and wave physics, and is currently on leave from his position of Professor of Physics at the Université de Paris. Prior to that or in parallel, he has held various academic positions: fellow of the Institut Universitaire de France, associate professor at Universite Pierre et Marie Curie, Visiting Senior Lecturer at Queen Mary University of London, UK, Visiting Professor at the National Institute for Informatics in Tokyo, Japan. Laurent has authored or co-authored more than 200 scientific publications, has been a consultant to various small and large companies, and is a co-inventor in several patents. He is a graduate in physics from Ecole Normale Superieure in Paris, and holds a PhD in Applied Mathematics from Marseille University. *logistics*: Participants are invited to follow the seminar either online or in person at: Room Turing Conseil, 7th floor (make a right on exiting the elevators and left at the end of the corridor), 45 rue des Saints Pères >; visitors should have an ID with them. Pour recevoir les annonces du diiP, vous pouvez vous inscrire à notre mailing list en envoyant un mail à diip@math-info.univ-paris5.fr  avec « follow diiP » dans l'objet du mail. Bien à vous, Anaïs pour le diiP Anaïs de Muret Manager de projets, Cité du Genre  et Data Intelligence Institute of Paris Université de Paris 5, rue Thomas Mann Bâtiment C des Grands Moulins - Bureau 879C 75013 Paris 01 57 27 52 36 --------------eie7gDzqhJJivPBUNcS0Qn8J Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: 8bit

diiP Distinguished Lecture:

Promises and challenges of massive-scale AI – the case of large language models



Who: Laurent Daudet, CTO and co-founder at LightOn, Professor (on leave) of physics (Université de Paris)

When: November 3, 4pm (Paris time).

Where: hybrid: online (zoom) and at Room Turing Conseil, 45 rue des Saints Pères 75006 Paris

zoom link: <https://u-paris.zoom.us/j/81455236254?pwd=Y0lHNkNpSUd6UUFjd2xSQW5wSlVMdz09>


title: Promises and challenges of massive-scale AI – the case of large language models


abstract: OpenAi’s GPT-3 language model has triggered a new generation of Machine Learning models. Leveraging Transformers architectures at billion-size parameters trained on massive unlabeled datasets, these language models achieve new capabilities such as text generation, question answering, or even zero-shot learning - tasks the model has not been explicitly trained for. However, training these models represent massive computing tasks, now done on dedicated supercomputers. Scaling up these models will require new hardware and optimized training algorithms. 


At LightOn - a spinoff of university research -, we develop a set of technologies to address these challenges. The Optical Processing Unit (OPU) technology makes some matrix-vector multiplications in a massively parallel fashion, at record-low power consumption. Now accessible on-premises or through the cloud, the OPU technology has been used by engineers and researchers worldwide in a variety of applications, for Machine Learning and scientific computing. We also train in an efficient manner large language models, such as PAGnol (demo at https://pagnol.lighton.ai ), the largest language model in French, that can be used for various research and business applications. 


short bio: Laurent Daudet is currently employed as CTO at LightOn, a startup he co-founded in 2016, where he manages cross-disciplinary R&D projects, involving machine learning, optics, signal processing, electronics, and software engineering. Laurent is a recognized expert in signal processing and wave physics, and is currently on leave from his position of Professor of Physics at the Université de Paris. Prior to that or in parallel, he has held various academic positions: fellow of the Institut Universitaire de France, associate professor at Universite Pierre et Marie Curie, Visiting Senior Lecturer at Queen Mary University of London, UK, Visiting Professor at the National Institute for Informatics in Tokyo, Japan. Laurent has authored or co-authored more than 200 scientific publications, has been a consultant to various small and large companies, and is a co-inventor in several patents. He is a graduate in physics from Ecole Normale Superieure in Paris, and holds a PhD in Applied Mathematics from Marseille University.


logistics:

Participants are invited to follow the seminar either online or in person at: Room Turing Conseil, 7th floor (make a right on exiting the elevators and left at the end of the corridor), 45 rue des Saints Pères <http://lipade.mi.parisdescartes.fr/?page_id=89>; visitors should have an ID with them. 



Pour recevoir les annonces du diiP, vous pouvez vous inscrire à notre mailing list en envoyant un mail à diip@math-info.univ-paris5.fr avec « follow diiP » dans l'objet du mail.


Bien à vous,

Anaïs pour le diiP


Anaïs de Muret

Manager de projets, Cité du Genre et Data Intelligence Institute of Paris

Université de Paris


5, rue Thomas Mann

Bâtiment C des Grands Moulins - Bureau 879C

75013 Paris


01 57 27 52 36 

--------------eie7gDzqhJJivPBUNcS0Qn8J--