import os

BASE_DIR = "/app"
HF_DATASET_REPO = "Zubaish/gandhi-kb-docx"  # Ensure this points to your NEW docx repo
HF_TOKEN = os.getenv("HF_TOKEN")

CHROMA_DIR = os.path.join(BASE_DIR, "chroma_db")
KB_DIR = os.path.join(BASE_DIR, "kb")

EMBEDDING_MODEL = "sentence-transformers/all-MiniLM-L6-v2"
LLM_MODEL = "Qwen/Qwen2.5-1.5B-Instruct" 
LLM_TASK = "text-generation"

CHUNK_SIZE = 1000
CHUNK_OVERLAP = 100