Rerun a model training

It is possible to relaunch the training of a model, whether this training terminated succesfully or with an error.

By default, the model training form is pre-filled with the previously set parameters: parent model, domain and Datasets. If you’ve made changes to the selected Dataset in the mean time, the modifications will be taken into account in the new training. For instance, if you added a new corpus, it will be used during the new training.


It is not possible to retrain a model with an expired Dataset. Please create a new Dataset.

There are two ways to relaunch a training.

  • Tick the box next to the model, then click on the Rerun training button:

  • Click on the cog next to the model, and choose Rerun training from the drop-down menu:

You can edit the pre-filled values, corresponding to the parameters set for the previous model (i.e. the model you initially selected). For instance, as shown in the example below, the dataset used for the initial training is already selected. All the values can be modified. You can also change your training data selection. Please refer to Create a new model for detailed information on the form fields.

  • Click on Submit to launch your new training.

You can only select one training to rerun at once. If you wish to rerun several trainings, please repeat the procedure for each model training.