voidful

/

Llama-Typhoon-8B-R1

like 9

I notice a majority amount of responses from this model doesn't include the hidden thought process , do you think its necessary to include <|start_header_id|>assistant<|end_header_id|><think>\\n instead of just <|start_header_id|>assistant<|end_header_id|> for generation prompt?