Backdoor detection through Activation Clustering **Unzip the art folder. To run experiment: python poison_defense_activation_clustering.py --model_name_or_path "bert-base-uncased" --data_path "datasets/sst2" --save_model_path "results" --poison_rate .5 --attack_type "badnet"