How to get sentence embedding using a fine-tuned model

I have fine-tuned a bert model using this script – transformers/run_glue.py at main · huggingface/transformers · GitHub

I was wondering how I can use the fine-tuned model to get sentence embedding.

Please suggest.