syntetic_issue_report_data_generation/config.py

from pathlib import Path

from dotenv import load_dotenv

# Load environment variables from .env file if it exists
load_dotenv()

# DIRECTORY PATHS
PROJ_ROOT = Path(__file__).resolve().parents[1]

DATA_DIR = PROJ_ROOT / "data"
RAW_DATA_DIR = DATA_DIR / "raw"
INTERIM_DATA_DIR = DATA_DIR / "interim"
ISSUE_REPORT_DIR = INTERIM_DATA_DIR / "issue-report-classification"
SOFT_CLEANED_DATA_DIR = ISSUE_REPORT_DIR / "soft-cleaned"
PROCESSED_DATA_DIR = DATA_DIR / "processed"
EXTERNAL_DATA_DIR = DATA_DIR / "external"
SAMPLES_DIR = DATA_DIR / "samples"
EMBEDDING_DIR = DATA_DIR / "embeddings"

MODELS_DIR = PROJ_ROOT / "models"

REPORTS_DIR = PROJ_ROOT / "reports"
FIGURES_DIR = REPORTS_DIR / "figures"

RANDOM_SEED = 42

DATASET_CONFIGs = {
    "nasa_cfs_train": {
        "data_path": "nasa/cfs_train.csv",
        "label_col": "label",
        "title_col": None,
        "body_col": "issue",
    },
    "nasa_fprime_train": {
        "data_path": "nasa/fprime_train.csv",
        "label_col": "label",
        "title_col": "title",
        "body_col": "body",
    },
    "nasa_train": {
        "data_path": "nasa/nasa_train_sample.csv",
        "label_col": "label",
        "title_col": None,
        "body_col": "text",
    },
    "nlbse23_train": {
        "data_path": "nlbse23/nlbse23-issue-classification-train.csv",
        "label_col": "labels",
        "title_col": "title",
        "body_col": "body",
    },
    "nlbse24_train": {
        "data_path": "nlbse24/issues_train.csv",
        "label_col": "label",
        "title_col": None,
        "body_col": "issue",
    },
    "pySenti4SD_train": {
        "data_path": "pySenti4SD/test_stackoverflow.csv",
        "label_col": "Polarity",
        "title_col": None,
        "body_col": "Text",
        "sep": ";",
    },
    "nasa_cfs_test": {
        "data_path": "nasa/cfs_test.csv",
        "label_col": "label",
        "title_col": None,
        "body_col": "issue",
    },
    "nasa_fprime_test": {
        "data_path": "nasa/fprime_test.csv",
        "label_col": "label",
        "title_col": "title",
        "body_col": "body",
    },
    "nasa_test": {
        "data_path": "nasa/nasa_test_sample.csv",
        "label_col": "label",
        "title_col": None,
        "body_col": "text",
    },
    "nlbse23_test": {
        "data_path": "nlbse23/nlbse23-issue-classification-test.csv",
        "label_col": "labels",
        "title_col": "title",
        "body_col": "body",
    },
    "nlbse24_test": {
        "data_path": "nlbse24/issues_test.csv",
        "label_col": "label",
        "title_col": None,
        "body_col": "issue",
    },
    "pySenti4SD_test": {
        "data_path": "pySenti4SD/test_stackoverflow.csv",
        "label_col": "Polarity",
        "title_col": None,
        "body_col": "Text",
        "sep": ";",
    },
    "test": {
        "data_path": "test/test.csv",
        "label_col": "label",
        "title_col": "title",
        "body_col": "body",
    },
}

# MODELS CONFIGURATION
MODEL_CONFIGS = {
    # SetFit Models
    "setfit-minilm": {
        "model_checkpoint": "sentence-transformers/all-MiniLM-L6-v2",
        "params": {
            "batch_size": 16,
            "num_epochs": 1,
            "num_iterations": 20,
            "learning_rate": 2e-5,
        },
    },
    "setfit-distilroberta": {
        "model_checkpoint": "sentence-transformers/all-distilroberta-v1",
        "params": {
            "batch_size": 16,
            "num_epochs": 1,
            "num_iterations": 20,
            "learning_rate": 2e-5,
        },
    },
    # Standard Transformers Models
    "modernbert-base": {
        "model_checkpoint": "answerdotai/ModernBERT-base",
        "params": {
            "per_device_train_batch_size": 16,
            "per_device_eval_batch_size": 32,
            "gradient_accumulation_steps": 4,
            "num_train_epochs": 10,
            "learning_rate": 2e-5,
            "weight_decay": 0.01,
            "warmup_steps": 500,
        },
    },
    "roberta-base": {
        "model_checkpoint": "roberta-base",
        "params": {
            "per_device_train_batch_size": 16,
            "per_device_eval_batch_size": 32,
            "gradient_accumulation_steps": 4,
            "num_train_epochs": 15,
            "learning_rate": 2e-5,
            "weight_decay": 0.01,
            "warmup_steps": 500,
        },
    },
}

# --- IMPOSTAZIONI MLFLOW ---
MLFLOW_TRACKING_URI = "https://dagshub.com/se4ai2526-uniba/Capibara.mlflow"
MLFLOW_EXPERIMENT_NAME = "Baselines_SetFit"