constants

Constants used throughout the project.

Accelerators

Bases: StrEnum

Enum for accelerator types.

Attributes:

Name	Type	Description
AUTO	str	Represents the automatic selection of accelerator based on availability.
CPU	str	Represents the Central Processing Unit as the accelerator.
GPU	str	Represents the Graphics Processing Unit as the accelerator.

Source code in src/configs/constants.py

class Accelerators(StrEnum):
    """
    Enum for accelerator types.

    Attributes:
        AUTO (str): Represents the automatic selection of accelerator based on availability.
        CPU (str): Represents the Central Processing Unit as the accelerator.
        GPU (str): Represents the Graphics Processing Unit as the accelerator.
    """

    AUTO = 'auto'
    CPU = 'cpu'
    GPU = 'gpu'

BackboneNames

Bases: StrEnum

Enum for backbone names.

Attributes:

Name	Type	Description
ROBERTA_BASE	str	Represents the base version of the RoBERTa model.
ROBERTA_LARGE	str	Represents the large version of the RoBERTa model.
ROBERTA_RACE	str	Represents the fine-tuned-on-RACE RoBERTa model.

Source code in src/configs/constants.py

class BackboneNames(StrEnum):
    """
    Enum for backbone names.

    Attributes:
        ROBERTA_BASE (str): Represents the base version of the RoBERTa model.
        ROBERTA_LARGE (str): Represents the large version of the RoBERTa model.
        ROBERTA_RACE (str): Represents the fine-tuned-on-RACE RoBERTa model.
    """

    ROBERTA_BASE = 'roberta-base'
    ROBERTA_LARGE = 'roberta-large'
    ROBERTA_RACE = 'LIAMF-USP/roberta-large-finetuned-race'
    XLM_ROBERTA_BASE = 'FacebookAI/xlm-roberta-base'
    XLM_ROBERTA_LARGE = 'FacebookAI/xlm-roberta-large'

ConfigName

Bases: StrEnum

Enum for config names.

Attributes:

Name	Type	Description
DATA	str	Represents the data config.
TRAINER	str	Represents the trainer config.
MODEL	str	Represents the model config.

Source code in src/configs/constants.py

class ConfigName(StrEnum):
    """
    Enum for config names.

    Attributes:
        DATA (str): Represents the data config.
        TRAINER (str): Represents the trainer config.
        MODEL (str): Represents the model config.
    """

    DATA = 'data'
    TRAINER = 'trainer'
    MODEL = 'model'

DLModelNames

Bases: StrEnum

Enum for model names.

Attributes:

Name	Type	Description
MAG_MODEL	str	Represents the name of the MAG model.
ROBERTEYE_MODEL	str	Represents the name of the Eye BERT model.
AHN_CNN_MODEL	str	Represents the name of the AHN CNN model.
AHN_RNN_MODEL	str	Represents the name of the AHN RNN model.
BEYELSTM_MODEL	str	Represents the name of the BEYELSTM model.
POSTFUSION_MODEL	str	Represents the name of the PostFusion model.
PLMAS_MODEL	str	Represents the name of the PLMAS model.
PLMASF_MODEL	str	Represents the name of the PLMASF model.

Source code in src/configs/constants.py

class DLModelNames(StrEnum):
    """
    Enum for model names.

    Attributes:
        MAG_MODEL (str): Represents the name of the MAG model.
        ROBERTEYE_MODEL (str): Represents the name of the Eye BERT model.
        AHN_CNN_MODEL (str): Represents the name of the AHN CNN model.
        AHN_RNN_MODEL (str): Represents the name of the AHN RNN model.
        BEYELSTM_MODEL (str): Represents the name of the BEYELSTM model.
        POSTFUSION_MODEL (str): Represents the name of the PostFusion model.
        PLMAS_MODEL (str): Represents the name of the PLMAS model.
        PLMASF_MODEL (str): Represents the name of the PLMASF model.
    """

    ROBERTEYE_MODEL = 'Roberteye'
    POSTFUSION_MODEL = 'PostFusionModel'
    PLMAS_MODEL = 'PLMASModel'
    PLMASF_MODEL = 'PLMASFModel'
    MAG_MODEL = 'MAGModel'
    AHN_CNN_MODEL = 'AhnCNNModel'
    AHN_RNN_MODEL = 'AhnRNNModel'
    BEYELSTM_MODEL = 'BEyeLSTMModel'

DataSets

Bases: StrEnum

DataSets is an enumeration that represents different datasets used in the application.

Attributes:

Name	Type	Description
ONESTOP	str	Represents the OneStop dataset.
COPCO	str	Represents the CopCo dataset.
POTEC	str	Represents the PoTeC dataset.
SBSAT	str	Represents the SBSAT dataset.
HALLUCINATION	str	Represents the IITBHGC dataset.
MECO_L2	str	Represents the MECO L2 dataset.
MECO_L2W1	str	Represents the MECO L2W1 dataset.
MECO_L2W2	str	Represents the MECO L2W2 dataset.

Source code in src/configs/constants.py

class DataSets(StrEnum):
    """
    DataSets is an enumeration that represents different datasets used in the application.

    Attributes:
        ONESTOP (str): Represents the OneStop dataset.
        COPCO (str): Represents the CopCo dataset.
        POTEC (str): Represents the PoTeC dataset.
        SBSAT (str): Represents the SBSAT dataset.
        HALLUCINATION (str): Represents the IITBHGC dataset.
        MECO_L2 (str): Represents the MECO L2 dataset.
        MECO_L2W1 (str): Represents the MECO L2W1 dataset.
        MECO_L2W2 (str): Represents the MECO L2W2 dataset.
    """

    ONESTOP = 'OneStop'
    COPCO = 'CopCo'
    POTEC = 'PoTeC'
    SBSAT = 'SBSAT'
    HALLUCINATION = 'IITBHGC'
    MECO_L2 = 'MECOL2'
    MECO_L2W1 = 'MECOL2W1'
    MECO_L2W2 = 'MECOL2W2'

DataType

Bases: StrEnum

DataType is an enumeration that represents different types of data used in the application.

Attributes:

Name	Type	Description
IA	str	Represents interest area data.
FIXATIONS	str	Represents fixation data.
RAW	str	Represents raw eye-tracking data.
TRIAL_LEVEL	str	Represents trial-level aggregated data.
METADATA	str	Represents metadata.

Source code in src/configs/constants.py

class DataType(StrEnum):
    """
    DataType is an enumeration that represents different types of data used in the application.

    Attributes:
        IA (str): Represents interest area data.
        FIXATIONS (str): Represents fixation data.
        RAW (str): Represents raw eye-tracking data.
        TRIAL_LEVEL (str): Represents trial-level aggregated data.
        METADATA (str): Represents metadata.
    """

    IA = 'ia'
    FIXATIONS = 'fixations'
    RAW = 'raw'
    TRIAL_LEVEL = 'trial_level'
    METADATA = 'metadata'

DatasetLanguage

Bases: StrEnum

Enum for dataset languages.

Attributes:

Name	Type	Description
ENGLISH	str	Represents English language datasets.
GERMAN	str	Represents German language datasets.
DANISH	str	Represents Danish language datasets.

Source code in src/configs/constants.py

class DatasetLanguage(StrEnum):
    """
    Enum for dataset languages.

    Attributes:
        ENGLISH (str): Represents English language datasets.
        GERMAN (str): Represents German language datasets.
        DANISH (str): Represents Danish language datasets.
    """

    ENGLISH = 'English'
    GERMAN = 'German'
    DANISH = 'Danish'

FeatureMode

Bases: StrEnum

Enum for feature modes.

Attributes:

Name	Type	Description
EYES	str	Represents the eyes feature mode.
WORDS	str	Represents the words feature mode.
EYES_WORDS	str	Represents the combined eyes and words feature mode.

Source code in src/configs/constants.py

class FeatureMode(StrEnum):
    """
    Enum for feature modes.

    Attributes:
        EYES (str): Represents the eyes feature mode.
        WORDS (str): Represents the words feature mode.
        EYES_WORDS (str): Represents the combined eyes and words feature mode.
    """

    EYES = 'eyes'
    WORDS = 'words'
    EYES_WORDS = 'eyes_words'

Fields

Bases: StrEnum

Enum for field names in the data.

Attributes:

Name	Type	Description
BATCH	str	Represents the article_batch.
PARAGRAPH_ID	str	Represents the paragraph_id.
UNIQUE_PARAGRAPH_ID	str	Represents the unique_paragraph_id.
ARTICLE_ID	str	Represents the article_id.
ARTICLE_IND	str	Represents the article_index.
QUESTION	str	Represents the question.
LEVEL	str	Represents the difficulty_level.
LIST	str	Represents the list_number.
PARAGRAPH	str	Represents the paragraph.
HAS_PREVIEW	str	Represents the question_preview.
SUBJECT_ID	str	Represents the participant_id.
FINAL_ANSWER	str	Represents the selected_answer_position.
REREAD	str	Represents the repeated_reading_trial.
IA_DATA_IA_ID_COL_NAME	str	Represents the IA_ID.
FIXATION_REPORT_IA_ID_COL_NAME	str	Represents the CURRENT_FIX_INTEREST_AREA_INDEX.
IS_CORRECT	str	Represents the is_correct.
PRACTICE	str	Represents the practice_trial.

Source code in src/configs/constants.py

class Fields(StrEnum):
    """
    Enum for field names in the data.

    Attributes:
        BATCH (str): Represents the article_batch.
        PARAGRAPH_ID (str): Represents the paragraph_id.
        UNIQUE_PARAGRAPH_ID (str): Represents the unique_paragraph_id.
        ARTICLE_ID (str): Represents the article_id.
        ARTICLE_IND (str): Represents the article_index.
        QUESTION (str): Represents the question.
        LEVEL (str): Represents the difficulty_level.
        LIST (str): Represents the list_number.
        PARAGRAPH (str): Represents the paragraph.
        HAS_PREVIEW (str): Represents the question_preview.
        SUBJECT_ID (str): Represents the participant_id.
        FINAL_ANSWER (str): Represents the selected_answer_position.
        REREAD (str): Represents the repeated_reading_trial.
        IA_DATA_IA_ID_COL_NAME (str): Represents the IA_ID.
        FIXATION_REPORT_IA_ID_COL_NAME (str): Represents the CURRENT_FIX_INTEREST_AREA_INDEX.
        IS_CORRECT (str): Represents the is_correct.
        PRACTICE (str): Represents the practice_trial.
    """

    UNIQUE_TRIAL_ID = 'unique_trial_id'
    BATCH = 'article_batch'
    PARAGRAPH_ID = 'paragraph_id'
    UNIQUE_PARAGRAPH_ID = 'unique_paragraph_id'
    ARTICLE_ID = 'article_id'
    LEVEL = 'difficulty_level'
    LIST = 'list_number'
    PARAGRAPH = 'paragraph'
    HAS_PREVIEW = 'question_preview'
    SUBJECT_ID = 'participant_id'
    REREAD = 'repeated_reading_trial'
    IA_DATA_IA_ID_COL_NAME = 'IA_ID'
    FIXATION_REPORT_IA_ID_COL_NAME = 'CURRENT_FIX_INTEREST_AREA_INDEX'
    IS_CORRECT = 'is_correct'
    PRACTICE = 'practice_trial'
    QUESTION = 'question'

ItemLevelFeaturesModes

Bases: StrEnum

Enum for item-level feature modes.

Attributes:

Source code in src/configs/constants.py

class ItemLevelFeaturesModes(StrEnum):
    """
    Enum for item-level feature modes.

    Attributes:
    """

    RF = 'RF'
    BEYELSTM = 'BEYELSTM'
    SVM = 'SVM'
    LOGISTIC = 'LOGISTIC'
    READING_SPEED = 'READING_SPEED'

MLModelNames

Bases: StrEnum

Enum for ML model names.

Attributes:

Name	Type	Description
LOGISTIC_REGRESSION	str	Represents the logistic regression model.
SVM	str	Represents the support vector machine model.
RANDOM_FOREST	str	Represents the random forest model.
DUMMY_CLASSIFIER	str	Represents the dummy classifier model.
XGBOOST	str	Represents the XGBoost model.
LOGISTIC_REGRESSION_REG	str	Represents the logistic regression regressor.
SVM_REG	str	Represents the support vector regressor.
RANDOM_FOREST_REG	str	Represents the random forest regressor.
DUMMY_REGRESSOR	str	Represents the dummy regressor.
XGBOOST_REG	str	Represents the XGBoost regressor.

Source code in src/configs/constants.py

class MLModelNames(StrEnum):
    """
    Enum for ML model names.

    Attributes:
        LOGISTIC_REGRESSION (str): Represents the logistic regression model.
        SVM (str): Represents the support vector machine model.
        RANDOM_FOREST (str): Represents the random forest model.
        DUMMY_CLASSIFIER (str): Represents the dummy classifier model.
        XGBOOST (str): Represents the XGBoost model.
        LOGISTIC_REGRESSION_REG (str): Represents the logistic regression regressor.
        SVM_REG (str): Represents the support vector regressor.
        RANDOM_FOREST_REG (str): Represents the random forest regressor.
        DUMMY_REGRESSOR (str): Represents the dummy regressor.
        XGBOOST_REG (str): Represents the XGBoost regressor.
    """

    LOGISTIC_REGRESSION = 'LogisticRegressionMLModel'
    SVM = 'SupportVectorMachineMLModel'
    RANDOM_FOREST = 'RandomForestMLModel'
    DUMMY_CLASSIFIER = 'DummyClassifierMLModel'
    XGBOOST = 'XGBoostMLModel'
    LINEAR_REG = 'LinearRegressionRegressorMLModel'
    SVM_REG = 'SupportVectorRegressorMLModel'
    RANDOM_FOREST_REG = 'RandomForestRegressorMLModel'
    DUMMY_REGRESSOR = 'DummyRegressorMLModel'
    XGBOOST_REG = 'XGBoostRegressorMLModel'

MatmulPrecisionLevel

Bases: StrEnum

Enum for matrix multiplication precision levels.

Attributes:

Name	Type	Description
HIGHEST	str	Corresponds to "highest".
HIGH	str	Corresponds to "high".
MEDIUM	str	Corresponds to "medium".

Source code in src/configs/constants.py

class MatmulPrecisionLevel(StrEnum):
    """
    Enum for matrix multiplication precision levels.

    Attributes:
        HIGHEST (str): Corresponds to "highest".
        HIGH (str): Corresponds to "high".
        MEDIUM (str): Corresponds to "medium".
    """

    HIGHEST = 'highest'
    HIGH = 'high'
    MEDIUM = 'medium'

NormalizationModes

Bases: StrEnum

Enum for normalization modes.

Attributes:

Name	Type	Description
ALL	str	Represents the mode where data is normalized based on all trials.
TRIAL	str	Represents the mode where data is normalized based on a trial level.
NONE	str	Represents the mode where no data is normalized.

Source code in src/configs/constants.py

class NormalizationModes(StrEnum):
    """
    Enum for normalization modes.

    Attributes:
        ALL (str): Represents the mode where data is normalized based on all trials.
        TRIAL (str): Represents the mode where data is normalized based on a trial level.
        NONE (str): Represents the mode where no data is normalized.
    """

    ALL = 'all'
    TRIAL = 'trial'
    NONE = 'none'

Precision

Bases: StrEnum

Enum for precision types.

Attributes:

Name	Type	Description
SIXTEEN_MIXED	str	Corresponds to "16-mixed".
THIRTY_TWO_TRUE	str	Corresponds to "32-true".

Source code in src/configs/constants.py

class Precision(StrEnum):
    """
    Enum for precision types.

    Attributes:
        SIXTEEN_MIXED (str): Corresponds to "16-mixed".
        THIRTY_TWO_TRUE (str): Corresponds to "32-true".
    """

    SIXTEEN_MIXED = '16-mixed'
    THIRTY_TWO_TRUE = '32-true'

PredMode

Bases: StrEnum

Enum for prediction modes.

Source code in src/configs/constants.py

class PredMode(StrEnum):
    """
    Enum for prediction modes.
    """

    RCS = 'RCS'  # Reading Comprehension Skill
    RC = 'RC'  # Reading Comprehension
    STD = 'STD'  # Subjective Text Difficulty
    TYP = 'TYP'  # Typicality
    CV = 'CV'  # Claim Verification
    LEX = 'LEX'  # Vocabulary Knowledge
    DE = 'DE'  # Domain Expertise

RunModes

Bases: StrEnum

Enum for run modes.

Attributes:

Name	Type	Description
DEBUG	str	Represents the debug mode used for debugging the code.
FAST_DEV_RUN	str	Represents the fast development run mode used for quick testing.
TRAIN	str	Represents the train mode used for training the model.

Source code in src/configs/constants.py

class RunModes(StrEnum):
    """
    Enum for run modes.

    Attributes:
        DEBUG (str): Represents the debug mode used for debugging the code.
        FAST_DEV_RUN (str): Represents the fast development run mode used for quick testing.
        TRAIN (str): Represents the train mode used for training the model.
    """

    DEBUG = 'debug'
    FAST_DEV_RUN = 'fast_dev_run'
    TRAIN = 'train'

Scaler

Bases: Enum

Enum for scaler types. Each scaler type is associated with a corresponding scaler class from sklearn.preprocessing.

Attributes:

Name	Type	Description
MIN_MAX_SCALER	type	Corresponds to sklearn.preprocessing.MinMaxScaler.
ROBUST_SCALER	type	Corresponds to sklearn.preprocessing.RobustScaler.
STANDARD_SCALER	type	Corresponds to sklearn.preprocessing.StandardScaler.

Source code in src/configs/constants.py

class Scaler(Enum):
    """
    Enum for scaler types. Each scaler type is associated with a
        corresponding scaler class from sklearn.preprocessing.

    Attributes:
        MIN_MAX_SCALER (type): Corresponds to sklearn.preprocessing.MinMaxScaler.
        ROBUST_SCALER (type): Corresponds to sklearn.preprocessing.RobustScaler.
        STANDARD_SCALER (type): Corresponds to sklearn.preprocessing.StandardScaler.
    """

    MIN_MAX_SCALER = MinMaxScaler
    ROBUST_SCALER = RobustScaler
    STANDARD_SCALER = StandardScaler

SetNames

Bases: StrEnum

Enum for set names.

Attributes:

Name	Type	Description
TRAIN	str	Represents the training set.
VAL	str	Represents the validation set.
TEST	str	Represents the test set.
SEEN_SUBJECT_UNSEEN_ITEM	str	Represents the seen subject unseen item set.
UNSEEN_SUBJECT_SEEN_ITEM	str	Represents the unseen subject seen item set.
UNSEEN_SUBJECT_UNSEEN_ITEM	str	Represents the unseen subject unseen item set.

Source code in src/configs/constants.py

class SetNames(StrEnum):
    """
    Enum for set names.

    Attributes:
        TRAIN (str): Represents the training set.
        VAL (str): Represents the validation set.
        TEST (str): Represents the test set.
        SEEN_SUBJECT_UNSEEN_ITEM (str): Represents the seen subject unseen item set.
        UNSEEN_SUBJECT_SEEN_ITEM (str): Represents the unseen subject seen item set.
        UNSEEN_SUBJECT_UNSEEN_ITEM (str): Represents the unseen subject unseen item set.
    """

    TRAIN = 'train'
    VAL = 'val'
    TEST = 'test'
    SEEN_SUBJECT_UNSEEN_ITEM = 'seen_subject_unseen_item'
    UNSEEN_SUBJECT_SEEN_ITEM = 'unseen_subject_seen_item'
    UNSEEN_SUBJECT_UNSEEN_ITEM = 'unseen_subject_unseen_item'

TaskTypes

Bases: StrEnum

Enum for task types.

Attributes:

Name	Type	Description
BINARY_CLASSIFICATION	str	Represents binary classification tasks.
REGRESSION	str	Represents regression tasks.

Source code in src/configs/constants.py

class TaskTypes(StrEnum):
    """
    Enum for task types.

    Attributes:
        BINARY_CLASSIFICATION (str): Represents binary classification tasks.
        REGRESSION (str): Represents regression tasks.
    """

    BINARY_CLASSIFICATION = 'binary_classification'
    REGRESSION = 'regression'