cudabenchmarktest/qwen3.5-9b-qwen3.6-reasoning-distilled-GGUF

INVALID LANGUAGE PAIR SPECIFIED. EXAMPLE: LANGPAIR=EN|IT USING 2 LETTER ISO OR RFC3066 LIKE ZH-CN. ALMOST ALL LANGUAGES SUPPORTED BUT SOME MAY HAVE NO CONTENT On longer prompts the model can stay inside the block > past the response budget, never emit , and produce zero answer > tokens on 25-46% of requests. > > Set think: true (or omit) only when you DO want chain-of-thought reasoning > (math, planning, complex multi-step). This is Qwen3 dual-mode operation per > https://qwenlm.github.io/blog/qwen3/. > > See the dataset cudabenchmarktest/r9-research-framework > OLLAMAINFERENCEWARNING.md for the full explanation.

Модальности:
Генерация текста

Области применения:
Логика и рассуждение Диалог / чат

Задача: Генерация текста
Автор: cudabenchmarktest
Теги: gguf, qwen3.5, reasoning, distillation, lora, sft, endpoints_compatible, conversational
Лайков: 3  |  Загрузок: 150

Открыть на HuggingFace →

Описание основано на материалах HuggingFace. Перевод выполнен автоматически.