Create custom LLM evals

Enter your custom task prompt and example traces, get eval metrics, and refine them. Then, export your evals or graded data for fine-tuning.