Kette der Dichte-Zusammenfassungsaufforderung

[ { "Missing_Entities": "German translation; verbatim placeholders; structure preservation", "Denser_Summary": "This article discusses the task of translating content into German while preserving verbatim placeholders like [X] and {x}, brand names, parameter flags such as --ar and 85mm, and code or JSON keys. It emphasizes keeping the original structure exactly, avoiding em-dashes in favor of hyphens, and maintaining diacritics. The output must be only the translation without any additional text or formatting." }, { "Missing_Entities": "entity-dense summaries; iterative refinement; JSON output", "Denser_Summary": "This article discusses translating into German with verbatim placeholders [X]/{x}, brand names, flags (--ar,85mm,4K,24fps), code/JSON keys, and structure 1:1, using ' - ' not em-dashes, preserving diacritics. It also outlines generating concise, entity-dense summaries iteratively, identifying missing entities, and outputting a JSON list of five dictionaries with keys 'Missing_Entities' and 'Denser_Summary'." }, { "Missing_Entities": "5 steps; identical word count; fusion and compression", "Denser_Summary": "Translating into German preserves [X]/{x}, brands, flags (--ar,85mm,4K,24fps), JSON keys, structure 1:1, ' - ' for dashes, diacritics. Summaries are refined over 5 steps, each adding 1-3 missing entities, keeping identical word count, using fusion and compression to fit new details, and outputting JSON with 'Missing_Entities' and 'Denser_Summary'." }, { "Missing_Entities": "verbose first summary; ~80 words; self-contained", "Denser_Summary": "German translation preserves [X]/{x}, brands, flags (--ar,85mm,4K,24fps), JSON keys, structure 1:1, ' - ' not em-dashes, diacritics. Summaries evolve in 5 steps, adding 1-3 missing entities, same word count, via fusion/compression. First summary is verbose (~80 words), later ones dense and self-contained, output as JSON list of five dicts." }, { "Missing_Entities": "never drop entities; add fewer if needed; faithful to article", "Denser_Summary": "German translation preserves [X]/{x}, brands, flags (--ar,85mm,4K,24fps), JSON keys, structure 1:1, ' - ' not em-dashes, diacritics. Summaries refine over 5 steps, adding 1-3 missing entities, same word count, via fusion/compression. First is verbose (~80 words), later dense/self-contained. Never drop prior entities; if space tight, add fewer. Output JSON list of five dicts with 'Missing_Entities' and 'Denser_Summary'." } ]