Open Models Almanac
Methodsevaluationbenchmark

single mention · evaluation

Execution-grounded evaluation

Evaluation that runs model outputs in real software environments and compiles, executes, and tests them.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See benchmark for the methods filed beside it.

Filed under evaluation › benchmark