Closed ma-beaudoin closed 1 month ago
Currently, the output folder and the best_model inside it do not get copied in the tmp_folder on the cluster, therefore model training does not resume as expected. Proposed fix: copy the output_dir in tmp_folder at initialization.
solved in #122
mirkobronzi
commented
1 month ago
Currently, the output folder and the best_model inside it do not get copied in the tmp_folder on the cluster, therefore model training does not resume as expected. Proposed fix: copy the output_dir in tmp_folder at initialization.