inference.py

from sentence_transformers import SentenceTransformer

def load_model():
    return SentenceTransformer("./")

def encode(texts):
    model = load_model()
    return model.encode(texts, convert_to_numpy=True).tolist()