Module monk.gluon.finetune.level_5_state_base

Classes

class finetune_state (verbose=1)

Base class for Monk states - train, eval_infer, resume, copy_from, pseudo_copy_from (for running sub-experiments)

Args

verbose : int

Set verbosity levels 0 - Print Nothing 1 - Print desired details

Expand source code

class finetune_state(finetune_evaluation):
    '''
    Base class for Monk states - train, eval_infer, resume, copy_from, pseudo_copy_from (for running sub-experiments)

    Args:
        verbose (int): Set verbosity levels
                        0 - Print Nothing
                        1 - Print desired details
    '''

    def __init__(self, verbose=1):
        super().__init__(verbose=verbose);


    
    ###############################################################################################################################################
    def set_system_state_eval_infer(self):
        '''
        Set system for eval_infer state

        Args:
            None

        Returns:
            None
        '''
        self.system_dict = read_json(self.system_dict["fname"], verbose=self.system_dict["verbose"]);
        self.system_dict["states"]["eval_infer"] = True;

        if(self.system_dict["training"]["status"]):
            if(len(self.system_dict["dataset"]["transforms"]["test"])):
                self.system_dict = retrieve_test_transforms(self.system_dict);
            else:
                self.custom_print("Test transforms not found.");
                self.custom_print("Add test transforms");
                self.custom_print("");
            self.set_model_final();
        else:
            msg = "Model in {} not trained. Cannot perform testing or inferencing".format(self.system_dict["experiment_name"]);
            raise ConstraintError(msg);
    ###############################################################################################################################################


    ###############################################################################################################################################
    def set_system_state_resume_train(self):
        '''
        Set system for resume training state

        Args:
            None

        Returns:
            None
        '''
        self.system_dict = read_json(self.system_dict["fname"], verbose=self.system_dict["verbose"]);
        self.system_dict["states"]["resume_train"] = True;
        if(self.system_dict["dataset"]["status"]):
            self.system_dict = retrieve_trainval_transforms(self.system_dict);
            self.set_dataset_final();
        else:
            msg = "Dataset not set.\n";
            msg += "Training not started. Cannot Run resume Mode";
            raise ConstraintError(msg);
        if(self.system_dict["model"]["status"]):
            self.set_model_final();
        else:
            msg = "Model not set.\n";
            msg += "Training not started. Cannot Run resume Mode";
            raise ConstraintError(msg);
        if(self.system_dict["hyper-parameters"]["status"]):
            self.system_dict = retrieve_optimizer(self.system_dict);
            self.system_dict = retrieve_scheduler(self.system_dict);
            self.system_dict = retrieve_loss(self.system_dict);
        else:
            msg = "hyper-parameters not set.\n";
            msg += "Training not started. Cannot Run resume Mode";
            raise ConstraintError(msg);
    ###############################################################################################################################################


    ###############################################################################################################################################
    def set_system_state_copy_from(self, copy_from):
        '''
        Set system for copied state

        Args:
            None

        Returns:
            None
        '''
        fname = self.system_dict["master_systems_dir_relative"] + copy_from[0] + "/" + copy_from[1] + "/experiment_state.json";
        system_dict_tmp = read_json(fname, verbose=self.system_dict["verbose"]);

        if(not system_dict_tmp["training"]["status"]):
            self.custom_print("Project - {}, Experiment - {}, has incomplete training.".format(copy_from[0], copy_from[1]));
            self.custom_print("Complete Previous training before copying from it.");
            self.custom_print("");
        elif(copy_from[0] == self.system_dict["project_name"] and copy_from[1] == self.system_dict["experiment_name"]):
            self.custom_print("Cannot copy same experiment. Use a different experiment to copy and load a previous experiment");
            self.custom_print("");
        else:
            self.system_dict["dataset"] = system_dict_tmp["dataset"];
            self.system_dict["model"] = system_dict_tmp["model"];
            self.system_dict["hyper-parameters"] = system_dict_tmp["hyper-parameters"];
            self.system_dict["training"] = system_dict_tmp["training"];
            self.system_dict["origin"] = [copy_from[0], copy_from[1]];
            self.system_dict["training"]["outputs"] = {};
            self.system_dict["training"]["outputs"]["max_gpu_memory_usage"] = 0;
            self.system_dict["training"]["outputs"]["best_val_acc"] = 0;
            self.system_dict["training"]["outputs"]["best_val_acc_epoch_num"] = 0;
            self.system_dict["training"]["outputs"]["epochs_completed"] = 0;
            self.system_dict["training"]["status"] = False;
            self.system_dict["training"]["enabled"] = True;
            self.system_dict["testing"] = {};
            self.system_dict["testing"]["status"] = False;
            save(self.system_dict);

            self.system_dict = read_json(self.system_dict["fname_relative"], verbose=self.system_dict["verbose"]);
            self.system_dict["states"]["copy_from"] = True;
            self.system_dict = retrieve_trainval_transforms(self.system_dict);
            self.Dataset();
            self.set_model_final();
            self.system_dict = retrieve_optimizer(self.system_dict);
            self.system_dict = retrieve_scheduler(self.system_dict);
            self.system_dict = retrieve_loss(self.system_dict);
    ###############################################################################################################################################



    ###############################################################################################################################################
    def set_system_state_pseudo_copy_from(self, pseudo_copy_from):
        '''
        Set system for copied state in hyper-parameter analysis mode 

        Args:
            None

        Returns:
            None
        '''
        fname = self.system_dict["master_systems_dir_relative"] + pseudo_copy_from[0] + "/" + pseudo_copy_from[1] + "/experiment_state.json";
        system_dict_tmp = read_json(fname, verbose=self.system_dict["verbose"]);
        self.system_dict["dataset"] = system_dict_tmp["dataset"];
        self.system_dict["model"] = system_dict_tmp["model"];
        self.system_dict["hyper-parameters"] = system_dict_tmp["hyper-parameters"];
        self.system_dict["training"] = system_dict_tmp["training"];
        self.system_dict["origin"] = [pseudo_copy_from[0], pseudo_copy_from[1]];
        self.system_dict["training"]["outputs"] = {};
        self.system_dict["training"]["outputs"]["max_gpu_memory_usage"] = 0;
        self.system_dict["training"]["outputs"]["best_val_acc"] = 0;
        self.system_dict["training"]["outputs"]["best_val_acc_epoch_num"] = 0;
        self.system_dict["training"]["outputs"]["epochs_completed"] = 0;
        self.system_dict["training"]["status"] = False;
        self.system_dict["training"]["enabled"] = True;
        self.system_dict["testing"] = {};
        self.system_dict["testing"]["status"] = False;
        save(self.system_dict);

        self.system_dict = read_json(self.system_dict["fname_relative"], verbose=self.system_dict["verbose"]);
        self.system_dict["states"]["pseudo_copy_from"] = True;
        self.system_dict = retrieve_trainval_transforms(self.system_dict);
        self.Dataset();
        self.set_model_final();
        self.system_dict = retrieve_optimizer(self.system_dict);
        self.system_dict = retrieve_scheduler(self.system_dict);
        self.system_dict = retrieve_loss(self.system_dict);

Ancestors

• gluon.finetune.level_4_evaluation_base.finetune_evaluation
• gluon.finetune.level_3_training_base.finetune_training
• gluon.finetune.level_2_model_base.finetune_model
• gluon.finetune.level_1_dataset_base.finetune_dataset
• system.base_class.system

Methods

def set_system_state_copy_from(self, copy_from)

Set system for copied state

Args

None

 

Returns

None

 

Expand source code

def set_system_state_copy_from(self, copy_from):
    '''
    Set system for copied state

    Args:
        None

    Returns:
        None
    '''
    fname = self.system_dict["master_systems_dir_relative"] + copy_from[0] + "/" + copy_from[1] + "/experiment_state.json";
    system_dict_tmp = read_json(fname, verbose=self.system_dict["verbose"]);

    if(not system_dict_tmp["training"]["status"]):
        self.custom_print("Project - {}, Experiment - {}, has incomplete training.".format(copy_from[0], copy_from[1]));
        self.custom_print("Complete Previous training before copying from it.");
        self.custom_print("");
    elif(copy_from[0] == self.system_dict["project_name"] and copy_from[1] == self.system_dict["experiment_name"]):
        self.custom_print("Cannot copy same experiment. Use a different experiment to copy and load a previous experiment");
        self.custom_print("");
    else:
        self.system_dict["dataset"] = system_dict_tmp["dataset"];
        self.system_dict["model"] = system_dict_tmp["model"];
        self.system_dict["hyper-parameters"] = system_dict_tmp["hyper-parameters"];
        self.system_dict["training"] = system_dict_tmp["training"];
        self.system_dict["origin"] = [copy_from[0], copy_from[1]];
        self.system_dict["training"]["outputs"] = {};
        self.system_dict["training"]["outputs"]["max_gpu_memory_usage"] = 0;
        self.system_dict["training"]["outputs"]["best_val_acc"] = 0;
        self.system_dict["training"]["outputs"]["best_val_acc_epoch_num"] = 0;
        self.system_dict["training"]["outputs"]["epochs_completed"] = 0;
        self.system_dict["training"]["status"] = False;
        self.system_dict["training"]["enabled"] = True;
        self.system_dict["testing"] = {};
        self.system_dict["testing"]["status"] = False;
        save(self.system_dict);

        self.system_dict = read_json(self.system_dict["fname_relative"], verbose=self.system_dict["verbose"]);
        self.system_dict["states"]["copy_from"] = True;
        self.system_dict = retrieve_trainval_transforms(self.system_dict);
        self.Dataset();
        self.set_model_final();
        self.system_dict = retrieve_optimizer(self.system_dict);
        self.system_dict = retrieve_scheduler(self.system_dict);
        self.system_dict = retrieve_loss(self.system_dict);

def set_system_state_eval_infer(self)

Set system for eval_infer state

Args

None

 

Returns

None

 

Expand source code

def set_system_state_eval_infer(self):
    '''
    Set system for eval_infer state

    Args:
        None

    Returns:
        None
    '''
    self.system_dict = read_json(self.system_dict["fname"], verbose=self.system_dict["verbose"]);
    self.system_dict["states"]["eval_infer"] = True;

    if(self.system_dict["training"]["status"]):
        if(len(self.system_dict["dataset"]["transforms"]["test"])):
            self.system_dict = retrieve_test_transforms(self.system_dict);
        else:
            self.custom_print("Test transforms not found.");
            self.custom_print("Add test transforms");
            self.custom_print("");
        self.set_model_final();
    else:
        msg = "Model in {} not trained. Cannot perform testing or inferencing".format(self.system_dict["experiment_name"]);
        raise ConstraintError(msg);

def set_system_state_pseudo_copy_from(self, pseudo_copy_from)

Set system for copied state in hyper-parameter analysis mode

Args

None

 

Returns

None

 

Expand source code

def set_system_state_pseudo_copy_from(self, pseudo_copy_from):
    '''
    Set system for copied state in hyper-parameter analysis mode 

    Args:
        None

    Returns:
        None
    '''
    fname = self.system_dict["master_systems_dir_relative"] + pseudo_copy_from[0] + "/" + pseudo_copy_from[1] + "/experiment_state.json";
    system_dict_tmp = read_json(fname, verbose=self.system_dict["verbose"]);
    self.system_dict["dataset"] = system_dict_tmp["dataset"];
    self.system_dict["model"] = system_dict_tmp["model"];
    self.system_dict["hyper-parameters"] = system_dict_tmp["hyper-parameters"];
    self.system_dict["training"] = system_dict_tmp["training"];
    self.system_dict["origin"] = [pseudo_copy_from[0], pseudo_copy_from[1]];
    self.system_dict["training"]["outputs"] = {};
    self.system_dict["training"]["outputs"]["max_gpu_memory_usage"] = 0;
    self.system_dict["training"]["outputs"]["best_val_acc"] = 0;
    self.system_dict["training"]["outputs"]["best_val_acc_epoch_num"] = 0;
    self.system_dict["training"]["outputs"]["epochs_completed"] = 0;
    self.system_dict["training"]["status"] = False;
    self.system_dict["training"]["enabled"] = True;
    self.system_dict["testing"] = {};
    self.system_dict["testing"]["status"] = False;
    save(self.system_dict);

    self.system_dict = read_json(self.system_dict["fname_relative"], verbose=self.system_dict["verbose"]);
    self.system_dict["states"]["pseudo_copy_from"] = True;
    self.system_dict = retrieve_trainval_transforms(self.system_dict);
    self.Dataset();
    self.set_model_final();
    self.system_dict = retrieve_optimizer(self.system_dict);
    self.system_dict = retrieve_scheduler(self.system_dict);
    self.system_dict = retrieve_loss(self.system_dict);

def set_system_state_resume_train(self)

Set system for resume training state

Args

None

 

Returns

None

 

Expand source code

def set_system_state_resume_train(self):
    '''
    Set system for resume training state

    Args:
        None

    Returns:
        None
    '''
    self.system_dict = read_json(self.system_dict["fname"], verbose=self.system_dict["verbose"]);
    self.system_dict["states"]["resume_train"] = True;
    if(self.system_dict["dataset"]["status"]):
        self.system_dict = retrieve_trainval_transforms(self.system_dict);
        self.set_dataset_final();
    else:
        msg = "Dataset not set.\n";
        msg += "Training not started. Cannot Run resume Mode";
        raise ConstraintError(msg);
    if(self.system_dict["model"]["status"]):
        self.set_model_final();
    else:
        msg = "Model not set.\n";
        msg += "Training not started. Cannot Run resume Mode";
        raise ConstraintError(msg);
    if(self.system_dict["hyper-parameters"]["status"]):
        self.system_dict = retrieve_optimizer(self.system_dict);
        self.system_dict = retrieve_scheduler(self.system_dict);
        self.system_dict = retrieve_loss(self.system_dict);
    else:
        msg = "hyper-parameters not set.\n";
        msg += "Training not started. Cannot Run resume Mode";
        raise ConstraintError(msg);