app + benchmark

LM Evaluation Harness

A framework for evaluating language models across a broad collection of academic tasks.

What it is

The harness standardizes task definitions, model adapters, few-shot prompting, and metric calculation for model evaluation.

Related records