search

jolibrain / wheatley

Next-generation scheduling problem solver based on GNNs and Reinforcement Learning
Other
44 stars 11 forks source link

Provide an easy way to use a trained model for inference #76

Closed pierrot-lc closed 1 year ago

pierrot-lc commented 1 year ago

To do inference with a trained JSSP model, you can now use the script located at jssp/solve.py. Should solve #75 .

Exemple of args:

python3 -m jssp.solve\
    --path "./PATH/TO/EXPERIMENT/"\
    -- load_problem "./PATH/TO/EXPERIMENT.TXT"\
    [--first_machine_id_is_one]  # Optional, if you load taillard problems, you should tell that the index starts at one.

The schedule will be saved locally and the makespan will be printed on stdout.