DiaFill is a Japanese dialogue script generation model designed to produce natural, spoken-style dialogue scripts rich in fillers and brief utternaces. Unlike typical assistant models that respond to users, this model is fine-tuned to generate a multi-turn dialogue script between two speakers based on a given scenario (seed data). This model was fine-tuned on non-public Japanese dialogue script data created as part of the GENIAC (Generative AI Accelerator Challenge) project. The training data itself is not publicly available. The project is described in the following press release: https://www.softbank.jp/corp/news/press/sbkk/2025/20250213_01/ This model generates a dialogue script based on a «Seed» prompt describing the genre, topic, and speakers. — genre: Choose one of the following Japanese strings: 雑談 (Chit-chat) コールセンター（手続き） (Call Center — Procedure) コールセンター（テクニカルサポート） (Call Center — Technical Support) コールセンター（カスタマーサクセス） (Call Center — Customer Success) — industry: Required only for Call Center genres (e.g., «家電», «通信»). Omit this field if the genre is 雑談. — tone: Basically, choose either フランク (Casual) or 丁寧 (Polite) for each speaker.
Модальности:
Генерация текста
Области применения:
Диалог / чат Следование инструкциям
Задача: Генерация текста
Автор: sbintuitions
Теги: llama, dialogue-generation, spoken-dialogue, conversational, ja
Лайков: 4 | Загрузок: 74
Описание основано на материалах HuggingFace. Перевод выполнен автоматически.
