First, install the environment dependencies using environment.yml
-
Location of the main file for running the models :
DPL/src/dialogue_system/run/run.py
-
To use SD dataset : dataset_type = "SD" dataset_location = "src/classifier/data/sd_dataset"
-
To use MD dataset : dataset_type = "MD" dataset_location = "src/classifier/data/md_dataset"
-
To use DQN algorithm : dqn_type = "DQN"
-
To use Double DQN algorithm : dqn_type = "DoubleDQN"
-
To run Flat-DQN : agent_id = "agentdqn" disease_as_action = True use_all_labels = False
-
To run HRL : agent_id = "agenthrljoint2" allow_wrong_disease = False wrong_disease_knowledge = False sf_idf_knowledge = False disease_as_action = False classifier_type = "deep_learning" use_all_labels = True
-
To run DHER with only Trimming : is_trimming_used = True, is_stitching_used = False, agent_id = "agenthrljoint2", disease_as_action = False, classifier_type = "deep_learning"
-
To run DHER with only Stitching : is_trimming_used = False, is_stitching_used = True, agent_id = "agenthrljoint2", disease_as_action = False, classifier_type = "deep_learning",
To run DHER : is_trimming_used = True, is_stitching_used = True, agent_id = "agenthrljoint2", disease_as_action = False, classifier_type = "deep_learning"
For training : train_mode = True
For testing : train_mode = False, saved_model = model location (./../../../../src/model/DQN/checkpoint/)