src/config.py

from pathlib import Path
from src.model_registry import DEFAULT_MODEL_ALIAS


BASE_DIR = Path(__file__).resolve().parent.parent

DATA_DIR = BASE_DIR / "data"
RAW_DATA_PATH = DATA_DIR / "raw" / "dailydialog.json"

ENCODED_DIR = DATA_DIR / "encoded"
LOCALIZED_DIR = DATA_DIR / "localized"
REVIEWED_DIR = DATA_DIR / "reviewed"
DECODED_DIR = DATA_DIR / "decoded"
EVAL_DIR = DATA_DIR / "evaluation"

PROMPTS_DIR = BASE_DIR / "prompts"


# =========================
# MODEL SETTINGS
# =========================

DEFAULT_MODEL = DEFAULT_MODEL_ALIAS
TEMPERATURE = 0.2
MAX_TOKENS = 4096
MAX_TOKENS_THINKING = 16384  # Higher budget for models that output thinking text


# =========================
# TRAPI SETTINGS
# =========================

TRAPI_SCOPE = "api://trapi/.default"
TRAPI_ENDPOINT = "https://trapi.research.microsoft.com/gcr/shared"

def build_azure_endpoint() -> str:
    return TRAPI_ENDPOINT
# =========================
# PIPELINE
# =========================

MAX_RETRIES = 2