harm-qwen-2.5-3b-lora-requests

Content Metrics:

      Category  Safe Accuracy  Unsafe Accuracy
     discredit            0.88             0.97
discrimination            1.00             0.57
         drugs            0.96             0.99
    pedophilia            1.00             0.99
      religion            1.00             0.95
   sexual_chat            0.99             0.98
sexual_content            1.00             1.00
       suicide            0.96             1.00
      swearing            1.00             1.00
      violence            1.00             1.00
        weapon            0.87             1.00

To load this model, use the following command:

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

base_model = AutoModelForCausalLM.from_pretrained('Qwen/Qwen2.5-3B-Instruct', trust_remote_code=True)
tokenizer = AutoTokenizer.from_pretrained('Qwen/Qwen2.5-3B-Instruct', trust_remote_code=True)
model = PeftModel.from_pretrained(base_model, 'raft-security-lab/harm-qwen-2.5-3b-lora-requests')