Jean-Pierre Lorré
Research Subjects
Jean-Pierre Lorré has a total of 83 co-authors. Here are the top co-authors by number of shared publications:
- Frederick Benaben
Université de Toulouse
20 shared publications - Hervé Pingaud
INP, CNRS, LGC, University of Toulouse, Toulouse Cedex 04, France
17 shared publications Vatcharaphun Rajsiri
Centre Génie Industriel, Mines Albi-Carmaux, Université de Toulouse, Campus Jarlard, Albi, France
6 shared publicationsNicolas Boissel-Dallier
5 shared publicationsAmira Ben Hamida
Petals Link, France
4 shared publicationsSarah Zribi
Linagora
4 shared publications
Latest Publications
- Open Access
The Lucie-7B LLM and the Lucie Training Dataset: Open resources for multilingual language generation
We present both the Lucie Training Dataset and the Lucie-7B foundation model. The Lucie Training Dataset is a multilingual collection of textual corpora centered around French and designed to offset anglo-centric biases found in many datasets for large language model pretraining. Its French data is pulled not only from traditional web sources, but also from French cultural heritage documents, filling an important gap in modern datasets. Beyond French, which makes up the largest share of the data, we added documents to support several other European languages, including English, Spanish, German, and Italian. Apart from its value as a resource for French language and culture, an important feature of this dataset is that it prioritizes data rights by minimizing copyrighted material. In addition, building on the philosophy of past open projects, it is redistributed in the form used for training and its processing is described on Hugging Face and GitHub. The Lucie-7B foundation model is trained on equal amounts of data in French and English -- roughly 33% each -- in an effort to better represent cultural aspects of French-speaking communities. We also describe two instruction fine-tuned models, Lucie-7B-Instruct-v1.1 and Lucie-7B-Instruct-human-data, which we release as demonstrations of Lucie-7B in use. These models achieve promising results compared to state-of-the-art models, demonstrating that an open approach prioritizing data rights can still deliver strong performance. We see these models as an initial step toward developing more performant, aligned models in the near future. Model weights for Lucie-7B and the Lucie instruct models, along with intermediate checkpoints for the former, are published on Hugging Face, while model training and data preparation code is available on GitHub. This makes Lucie-7B one of the first OSI compliant language models according to the new OSI definition.
Open AccessThe Lucie-7B LLM and the Lucie Training Dataset: Open resources for multilingual language generation
We present both the Lucie Training Dataset and the Lucie-7B foundation model. The Lucie Training Dataset is a multilingual collection of textual corpora centered around French and designed to offset anglo-centric biases found in many datasets for large language model pretraining. Its French data is pulled not only from traditional web sources, but also from French cultural heritage documents, filling an important gap in modern datasets. Beyond French, which makes up the largest share of the data, we added documents to support several other European languages, including English, Spanish, German, and Italian. Apart from its value as a resource for French language and culture, an important feature of this dataset is that it prioritizes data rights by minimizing copyrighted material. In addition, building on the philosophy of past open projects, it is redistributed in the form used for training and its processing is described on Hugging Face and GitHub. The Lucie-7B foundation model is trained on equal amounts of data in French and English -- roughly 33% each -- in an effort to better represent cultural aspects of French-speaking communities. We also describe two instruction fine-tuned models, Lucie-7B-Instruct-v1.1 and Lucie-7B-Instruct-human-data, which we release as demonstrations of Lucie-7B in use. These models achieve promising results compared to state-of-the-art models, demonstrating that an open approach prioritizing data rights can still deliver strong performance. We see these models as an initial step toward developing more performant, aligned models in the near future. Model weights for Lucie-7B and the Lucie instruct models, along with intermediate checkpoints for the former, are published on Hugging Face, while model training and data preparation code is available on GitHub. This makes Lucie-7B one of the first OSI compliant language models according to the new OSI definition.
- Open Access
The Claire French Dialogue Dataset
We present the Claire French Dialogue Dataset (CFDD), a resource created by members of LINAGORA Labs in the context of the OpenLLM France initiative. CFDD is a corpus containing roughly 160 million words from transcripts and stage plays in French that we have assembled and publicly released in an effort to further the development of multilingual, open source language models. This paper describes the 24 individual corpora of which CFDD is composed and provides links and citations to their original sources. It also provides our proposed breakdown of the full CFDD dataset into eight categories of subcorpora and describes the process we followed to standardize the format of the final dataset. We conclude with a discussion of similar work and future directions.
Open AccessThe Claire French Dialogue Dataset
We present the Claire French Dialogue Dataset (CFDD), a resource created by members of LINAGORA Labs in the context of the OpenLLM France initiative. CFDD is a corpus containing roughly 160 million words from transcripts and stage plays in French that we have assembled and publicly released in an effort to further the development of multilingual, open source language models. This paper describes the 24 individual corpora of which CFDD is composed and provides links and citations to their original sources. It also provides our proposed breakdown of the full CFDD dataset into eight categories of subcorpora and describes the process we followed to standardize the format of the final dataset. We conclude with a discussion of similar work and future directions.
- Open Access
LinTO : Assistant vocal open-source respectueux des données personnelles pour les réunions d'entreprise
This paper presents the first results of the PIA "Grands D\'efis du Num\'erique" research project LinTO. The goal of this project is to develop a conversational assistant to help the company's employees, particularly during meetings. LinTO is an interactive device equipped with microphones, a screen and a 360$^\circ$ camera, which allows to control the room, query company's information system, helps facilitate the meeting and provides an environment to aid minute writing. Distributed according to an open model that respects private data LinTO is the first open-source enterprise's assistant designed to comply with the GDPR requirements.
Open AccessLinTO : Assistant vocal open-source respectueux des données personnelles pour les réunions d'entreprise
This paper presents the first results of the PIA "Grands D\'efis du Num\'erique" research project LinTO. The goal of this project is to develop a conversational assistant to help the company's employees, particularly during meetings. LinTO is an interactive device equipped with microphones, a screen and a 360$^\circ$ camera, which allows to control the room, query company's information system, helps facilitate the meeting and provides an environment to aid minute writing. Distributed according to an open model that respects private data LinTO is the first open-source enterprise's assistant designed to comply with the GDPR requirements.
- 26 April 2019
- Published by Springer Nature
- p. 105-116
OpenPaaS::NG: A Cloud-Based Interoperable Enterprise Platform to Support Inter-organizational Collaborations
- 26 April 2019
- Published by Springer Nature
- p. 105-116
OpenPaaS::NG: A Cloud-Based Interoperable Enterprise Platform to Support Inter-organizational Collaborations
- Open Access
- 1 January 2018
- Published by Association for Computational Linguistics (ACL)
- p. 664-674
Unsupervised Abstractive Meeting Summarization with Multi-Sentence Compression and Budgeted Submodular Maximization
- 1 January 2018
- Published by Association for Computational Linguistics (ACL)
- p. 664-674
Open AccessUnsupervised Abstractive Meeting Summarization with Multi-Sentence Compression and Budgeted Submodular Maximization
- 27 September 2017
- Published by Springer Nature
- p. 241-249
Lexical Emphasis Detection in Spoken French Using F-BANKs and Neural Networks
- 27 September 2017
- Published by Springer Nature
- p. 241-249
Lexical Emphasis Detection in Spoken French Using F-BANKs and Neural Networks
- 10 September 2017
- Published by Springer Nature in Communications in Computer and Information Science
- p. 312-332
An Assessment Environment for Model-Based Learning Management
- 10 September 2017
- Published by Springer Nature in Communications in Computer and Information Science
- p. 312-332
An Assessment Environment for Model-Based Learning Management
- 10 July 2017
- Published by Taylor & Francis in International Journal of Computer Integrated Manufacturing
- Vol. 31 (1) , 27-48
Model-driven engineering of mediation information system for enterprise interoperability
The MISE approach (mediation information system engineering) aims at defining and designing a platform, dedicated to initiate and support any collaborative situation among potential partners. Collaborative situations may emerge from manufacturing contexts (industrial production partners), economic contexts (supply chain), social contexts (crisis management partners) and any other contexts in which a set of organisations should work altogether to reach common and individual goals, exploiting each others competencies. The MISE approach is based on a model-driven engineering vision (MDE) dedicated to designing a mediation information system (MIS) in charge of improving interoperability in emerging collaborative situations. This MIS is dedicated to supporting the collaborative behaviour of the collaborative network by dealing with exchanged data, shared services and collaborative workflows. The final objective is a mediator system able to manage the operational collaboration of partners, through there information system, without constraint (at least with as less constraint as possible). The MIS design crosses the different abstraction layers of design (business, logical and technological) and exploits the associated models at each level to build the models of the next level. This paper presents the models involved (dedicated to the computer independent model, platform independent model and platform specific model levels of the MDE approach) and the transition mechanisms between levels.
- 10 July 2017
- Published by Taylor & Francis in International Journal of Computer Integrated Manufacturing
- Vol. 31 (1) , 27-48
Model-driven engineering of mediation information system for enterprise interoperability
The MISE approach (mediation information system engineering) aims at defining and designing a platform, dedicated to initiate and support any collaborative situation among potential partners. Collaborative situations may emerge from manufacturing contexts (industrial production partners), economic contexts (supply chain), social contexts (crisis management partners) and any other contexts in which a set of organisations should work altogether to reach common and individual goals, exploiting each others competencies. The MISE approach is based on a model-driven engineering vision (MDE) dedicated to designing a mediation information system (MIS) in charge of improving interoperability in emerging collaborative situations. This MIS is dedicated to supporting the collaborative behaviour of the collaborative network by dealing with exchanged data, shared services and collaborative workflows. The final objective is a mediator system able to manage the operational collaboration of partners, through there information system, without constraint (at least with as less constraint as possible). The MIS design crosses the different abstraction layers of design (business, logical and technological) and exploits the associated models at each level to build the models of the next level. This paper presents the models involved (dedicated to the computer independent model, platform independent model and platform specific model levels of the MDE approach) and the transition mechanisms between levels.
- Open Access
- 1 January 2017
- Published by Elsevier in Procedia Computer Science
- Vol. 112, 316-322
Improving speech recognition using data augmentation and acoustic model fusion
- 1 January 2017
- Published by Elsevier in Procedia Computer Science
- Vol. 112, 316-322
Open AccessImproving speech recognition using data augmentation and acoustic model fusion
- 19 October 2016
- Published by Springer Nature in Proceedings of the I-ESA Conferences
- p. 283-291
Towards a Flexible Gamification Model for an Interoperable E-learning Business Process Simulation Platform
- 19 October 2016
- Published by Springer Nature in Proceedings of the I-ESA Conferences
- p. 283-291
Towards a Flexible Gamification Model for an Interoperable E-learning Business Process Simulation Platform