← All work
Project / AI / Formal methods / Evaluation

AlloyEval

A course project replicating and extending an evaluation of LLM-generated Alloy specifications.

Full first page of Shmulik Cohen’s AlloyEval course report

AlloyEval studies how language models generate Alloy formal specifications. It reproduces and extends the public study by Hong et al. with additional task variants and a report.

The full report documents the task variants, experiments, and results.