MMLU & Language Model Evaluation

Objectives

• Understand benchmarks (MMLU and others) and their limits
• Build an evaluation set for your business
• Compare models on quality, cost and latency
• Evaluate continuously in production

Curriculum

1. Benchmark landscape3 h
2. MMLU: method and reading scores3 h
3. Building your own evaluation set4 h
4. Comparing quality, cost, latency2 h
5. Continuous evaluation2 h

Audience & prerequisites

Audience: Data scientists, AI engineers, CIOs and technical leads who select models.

Prerequisites: Python, machine learning basics.

Tools used

PythonEvaluation setsLLM APIs

Assessment and certification

The programme ends with a practical project assessed by a panel of practitioners. Passing it earns the Africa Data Entry certificate « MMLU & Language Model Evaluation ».