{"slug": "evaluating-prompt-scope-and-demonstration-similarity-in-local-llm-machine", "title": "Evaluating Prompt Scope and Demonstration Similarity in Local LLM Machine Translation", "summary": "A new arXiv study evaluating local LLM machine translation finds that dedicated MT systems from OPUS-MT and NLLB-200 remain stronger overall than three instruction-tuned LLMs (llama3.2:3b, mistral:latest, qwen2.5:14b) on English-to-Romance and English-to-Germanic translation across nine EU languages. Few-shot prompting helps mistral:latest and qwen2.5:14b but hurts llama3.2:3b, while family-scope prompting is feasible for stronger models but exposes failures in smaller ones.", "body_md": "arXiv:2607.26286v1 Announce Type: new\nAbstract: 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 sentence into one target language. In practice, users may ask for one target language, for several related languages at once, or for translations conditioned on examples. This paper studies prompt scope and demonstration selection as experimental variables for local LLM machine translation. We evaluate English-to-Romance and English-to-Germanic translation on the full FLORES devtest split for nine official European Union languages. We compare three local instruction-tuned LLMs, llama3.2:3b, mistral:latest, and qwen2.5:14b, against dedicated MT baselines from OPUS-MT and NLLB-200. We test zero-shot prompting and k=5 few-shot prompting with random, lexical-similarity, and embedding-similarity demonstration selection. We also compare single-target prompts with JSON-formatted family-scope prompts that request all languages in a family at once. Results show that dedicated MT systems remain strongest overall, especially for Germanic languages. Few-shot prompting helps mistral:latest and qwen2.5:14b, but hurts llama3.2:3b; embedding retrieval is best on average for the stronger LLMs, but its advantage over random and lexical examples is modest. Family-scope prompting is feasible for stronger local LLMs but exposes structured-output failures in smaller models. These findings motivate evaluating LLM translation not only by language pair and metric, but also by prompt scope, retrieval strategy, and multi-target compliance.", "url": "https://wpnews.pro/news/evaluating-prompt-scope-and-demonstration-similarity-in-local-llm-machine", "canonical_source": "https://arxiv.org/abs/2607.26286", "published_at": "2026-07-30 04:00:00+00:00", "updated_at": "2026-07-30 04:35:26.474172+00:00", "lang": "en", "topics": ["large-language-models", "machine-learning", "natural-language-processing"], "entities": ["arXiv", "OPUS-MT", "NLLB-200", "llama3.2:3b", "mistral:latest", "qwen2.5:14b"], "alternates": {"html": "https://wpnews.pro/news/evaluating-prompt-scope-and-demonstration-similarity-in-local-llm-machine", "markdown": "https://wpnews.pro/news/evaluating-prompt-scope-and-demonstration-similarity-in-local-llm-machine.md", "text": "https://wpnews.pro/news/evaluating-prompt-scope-and-demonstration-similarity-in-local-llm-machine.txt", "jsonld": "https://wpnews.pro/news/evaluating-prompt-scope-and-demonstration-similarity-in-local-llm-machine.jsonld"}}