Shiv Sakhuja’s Post

View profile for Shiv Sakhuja, graphic

Building an IDE for GenAI development | Co-founder, Athina AI (YC W23) | Ex-Google, YC

Mistral’s API client has an optional “safe_mode” flag, which helps guardrail the LLM for safety. They do this is by pre-pending this statement into your prompt. Will be interesting to see if more model providers bake in guardrails and safety nets into the API itself. Seems like a useful feature.

  • No alternative text description for this image
Tanu sri

AI Researcher LLMOps RAG | Passionate about Advancing AI Technologies

8mo

Honestly it just the prompt which is doing the magic everywhere. From analytics, classification to setting saftey boundaries, there no other complicated tech involved here.

Like
Reply

To view or add a comment, sign in

Explore topics