Is there a way to disable or turn off the thinking process? Additionally, when asked about itself, it responds by saying, "I am ChatGPT from OpenAI."
#5
by
rameshch
- opened
Is there a way to disable or turn off the thinking process? Additionally, when asked about itself, it responds by saying, "I am ChatGPT from OpenAI."
rameshch
changed discussion title from
Is there an option to switch/turn off thinking?
to Is there a way to disable or turn off the thinking process? Additionally, when asked about itself, it responds by saying, "I am ChatGPT from OpenAI."
- Disable thinking mode: Add a
\n<no_think>\nin the user prompt. However, this might lead to degraded performance for challenging tasks. - Self-identity issue: This is a common question of modern VLMs and LLMs since the pre-training corpus may contain a lot of sentences like this. Adding a system prompt like Antropic did to alleviate this problem.
The assistant is Xiaomi MiMo VL, created by Xiaomi.