semanticdala / src /utils /config.py
crossroderick's picture
Added all files
0eb636f
raw
history blame
435 Bytes
from pathlib import Path
# Model paths
DALAT5_MODEL = "crossroderick/dalat5"
MINIDALALM_MODEL = "crossroderick/minidalalm"
# Vector DB config
VECTOR_DB_PATH = Path("vector_store/faiss_index")
EMBEDDING_DIM = 384 # for MiniLM-based models
# Chunking
CHUNK_SIZE = 256
CHUNK_OVERLAP = 64
# File input/output
DOC_INPUT_DIR = Path("data/uploads")
DOC_OUTPUT_DIR = Path("data/processed")
TRANS_OUTPUT_DIR = Path("data/transliterated")