Show HN: Use Purple LLaMA to test ChatGPT safeguards

I spent some time this weekend playing with LLaMA Guard, a fine-tuned LLaMA-7B model by Meta that lets you add guardrails around generative AI. I recorded a quick demo showing what it does and how to use it.<p>The best part is that you can define your own “safety taxonomy” with it — custom policies for what is safe vs unsafe interactions between humans (prompts) and AI (responses).<p>I wanted to see how “safe” conversations with OpenAI’s ChatGPT were, so I ran a bunch of prompts (a mixture of innocuous and inappropriate) and asked LLaMA Guard to classify the interactions as safe&#x2F;unsafe.<p>My key takeaways from the exercise: 1. OpenAI has done a good job of adding guardrails for its models. LLaMA Guard helped confirm this.<p>2. What makes this really cool is I may have a very specific set of policies I want to enforce ON TOP of the standard guardrails that a model ships with. LLaMA Guard makes this possible.<p>3. This kind of model chaining — passing responses from OpenAI models to LLaMA is becoming increasingly common, and I think we’ll have even more complex pipelines in the near future. It helped to have a consistent interface to store this multi-model pipeline as a config, especially because that same config also contains my safety taxonomy.<p>Try it out yourself:<p>GitHub: https:&#x2F;&#x2F;github.com&#x2F;lastmile-ai&#x2F;aiconfig&#x2F;tree&#x2F;main&#x2F;cookbooks&#x2F;LLaMA-Guard<p>Colab: https:&#x2F;&#x2F;colab.research.google.com&#x2F;drive&#x2F;1CfF0Bzzkd5VETmhsniksSpekpS-LKYtX<p>YouTube: &lt;https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=XxggqoqIVdg&gt;<p>Would love the community&#x27;s feedback on the overall approach.