A research spike was done to evaluate a few different options for RAG eval frameworks based on the high-level needs of RAG Evals:

• Custom metrics (llm-based and non-llm)
• Custom evaluator models
• Integration with testing workflows
• Transparent evaluations

A meeting was then held to discuss the findings of this spike and to determine next steps. Here is recording of that meeting: https://drive.google.com/file/d/1yN3NPbkgMzjS7t17jHesiA7rFX0doj3F/view