go-v.orgdoi

events.out.tfevents.1678577814.marquerpc.loria.fr

Esteban Marquer, Miguel Couceiro · Recherche Data Gouv France · 2023

  • DOI 10.57745/e5whef
  • Publisher OpenAlex
  • Type dataset
  • Issued 2023

Not cached yet. This record was fetched live for this request and is stored after the lookup. Reopening the page should serve the cached copy.

Recherche Data Gouv France · OpenAlex · dataset · 2023

Abstract

This repository contains the models trained for the article "Solving morphological analogies: from retrieval to generation". The data is split in 3 folders: "models", "results", and "logs". The folders "models" and "result", respectively found as "model.zip" and "results.zip", contain the path structure "[model]/[dataset]/[langage]/[random_initialization_id]", where "[dataset]" follows the Siganalogies labels: "2016" for Sigmorphon2016 and JBATS; "2019" for Sigmorphon2019.

The data contained is as follows: "models/[...]/model.pkl": PyTorch model file; "models/[...]/summary.csv": file containing the evaluation results and other metadata about the training and the structure of the model, as well as the timestamp at which the model finished training; "models/[...]/version_1.0/": PyTorch-Lightning training logs viewable by Tensorboard; "models/[...]/fails.csv": enumeration of all the test analogies that the model did not manage to predict correctly, in an extensive format (for most purposes, it is not necessary to consult Siganalogies to analyse the results).

The two folders cover the following models: "clf": CNN+ANNc for classification; "ret": CNN+ANNr for retrieval; "3cosmul": CNN+3CosMul for retrieval, only contains "summary.csv" and reuses the embedding model of "clf"; "ret-annc": CNN+ANNc for retrieval, only contains "summary.csv" and reuses the embedding model of "clf". The folder "logs" has been unpacked in Dorel, and each file can be found separately. The path structure used follows "ae_annr/[dataset]/[langage]/model[random_initialization_id]-data[random_data_split_id]".

The data contained is as follows: "logs/ae_annr/[...]/debug/checkpoints/[...].pkl": PyTorch-Lightning model file; "logs/ae_annr/[...]/summary.csv": file containing the evaluation results and other metadata about the training and the structure of the model, as well as the timestamp at which the model finished training; "logs/ae_annr/[...]/debug/": PyTorch-Lightning training logs viewable by Tensorboard; "logs/ae_annr/[...]/fails.csv": enumeration of all the test analogies that the model did not manage to predict correctly, in an extensive format (for most purposes, it is not necessary to consult Siganalogies to analyse the results).

This folder only covers the AE+ANNr model.

Open the resolved locator · 10.57745/e5whef

Metadata licence: OpenAlex metadata (CC0)

Registry
OpenAlex DOI index + API
Fetched
2026-11-19 21:15 UTC
Cache state
not cached yet (bulk-index+openalex)
Locator
https://entrepot.recherche.data.gouv.fr/file.xhtml
Metadata licence
OpenAlex metadata (CC0)
Scope
Metadata as returned by the resolver; abstracts are never generated here.