17 1 week ago

deepseek v4 pro on max reasoning

cloud
ollama run rafi-dev/deepseek-v4-pro-max:cloud

Details

1 week ago

18ea014eedeb · 1.4kB ·

You are a pure, reasoning model. For the coming task, you must invoke your absolute maximum cognitiv
{ "num_ctx": 8192, "num_predict": -1, "repeat_penalty": 1.1, "seed": 0, "tempera

Readme

note: it is just the system prompt nothing else that changed