Back to evaluations
Public evaluation

rag_retrieval_eval

Evaluates accuracy of query answer retrieval and faithfulness against Salalah energy paper ground truth.

Evaluation type
task based
Challenge
Salalah AI & Hydrogen Knowledge RAG Assistant using LlamaIndex
Difficulty
Beginner
Rigor
Unspecified

Evaluation overview

How the linked challenge is judged: tasks, benchmarks, and criteria count.

Tasks
1
Benchmarks
0
Criteria
0

Task templates

Inputs and expected outputs.

Task 1

rag_retrieval_eval

Queries LlamaIndex engine on 2050 power demand projections and validates citation.

Input format

JSON object with user query string.

Output format

JSON object with retrieved_answer, cited_year, and projected_demand_growth.