semester project for 257-ml

poster

1. adv ( Advait Shinde )
2. jasss ( Jasper Morgal )
3. toneyyyyy ( Toney Zhen )
4. nickk ( Nickzad Bayati )

Guardrails:

• Input: Sensitive Data (Toney Zhen)
• Input: Prompt Injection (Advait Shinde)
• Output: General Malicious Output (Jasper Morgal)
• Output: Toxicity (Nickzad Bayati)

Output: Toxicity

1. download dataset from Kaggle: Jigsaw Toxic Comment Classification Challenge
2. unzip files and move them into /output_guardrails/toxicity/data/
3. use evaluate_model2.ipynb to train/evaluate models