Evaluating Prompt Scope and Demonstration Similarity in Local LLM Machine Translation

Mihael Arcan

Large language models (LLMs) are increasingly used as general-purpose translation systems, but their behavior is usually evaluated under a single prompt shape: translate one source…