Inventory of pre-trained learning models of the Etalab AI Lab

** Inventory of pre-trained learning models of the Etalab AI Lab** The publication of the inventory of pre-trained machine learning models is part of the roadmap of the Ministry of Transformation and Public Service (see p. 25 of the downloadable document [here](https://www.data.gouv.fr/fr/datasets/r/03b43dc4-b92f-4d40-9b7e-598dcd61c420)). This dataset lists the different algorithms trained by the Lab IA to date as part of the development of its shared tools (more information on the [dedicated page of the IA Lab](https://www.etalab.gouv.fr/outils-mutualises/)). **Details of what the inventory contains** For each algorithm, the column “link_model_card” provides a link to access a description of the algorithm. We followed the pattern description frame presented in Margaret Mitchell & al’s “Model Cards for Model Reporting” paper (downloadable [here](https://arxiv.org/abs/1810.03993)). The column “link_depot_github” returns to the GitHub repository containing the code that led to the algorithm. The column “model_entraine_open” has the value “no” if the trained model is not opened and is “yes” if the driven model is opened. In the latter case, the link to the driven model is entered in the column “link_modele_entraine_si_pertinent”. The column “date_last_mise_a_day” indicates the date of last update of the template.