Learning-Based Automated Adversarial Red-Teaming for Robustness Evaluation of Large Language Models