# script.py

from model2vec.distill import distill
import os

# Set local base model path (already cloned)
LOCAL_MODEL_PATH = os.path.join(os.path.dirname(__file__), "snowflake-arctic-embed-xs")
OUTPUT_PATH = os.path.dirname(os.path.abspath(__file__))
PCA_DIMS = 256  # You can tweak this

print("Distilling Model2Vec from local model...")

# Distill model
model = distill(model_name=LOCAL_MODEL_PATH, pca_dims=PCA_DIMS)

print("Saving distilled model to:", OUTPUT_PATH)
model.save_pretrained(OUTPUT_PATH)

print("✅ Done. Distilled model saved to", OUTPUT_PATH)