Study finds prompt engineering outweighs hyperparameter tuning for extracting data from Spanish electricity invoices using general-purpose LLMs

Researchers benchmarked Gemini 1.5 Pro and Mistral-small across 19 parameter configurations and 6 prompting strategies on a subset of the IDSEM dataset to extract structured information from Spanish electricity invoices without task-specific fine-tuning.

Prompt quality dominated performance: the F1-score variation across all parameter configurations was marginal, while the gap between zero-shot and the best few-shot strategy exceeded 19 percentage points. The best configuration (few-shot with cross-validation) achieved an F1-score of 97.61% for Gemini and 96.11% for Mistral-small.

Document template structure emerged as the primary determinant of extraction difficulty, establishing that prompt design is the critical lever for maximizing extraction fidelity in LLM-based document processing.