The community member has been experimenting with deepseek R1 on groq in typebot and found it to be generating output at a fast rate of about 275 tokens per second. They believe this speed would be sufficient to only show the output, without the chain of thought or thinking section. The community member suggests that it would be better if the groq integration module had an option to do this, rather than having to use a regex.
I've been playing around with deepseek R1 on groq in typebot and it works really well generating at about 275 tokens / second.
It's so fast that I think it would work well enough to only show the output (not the chain of thought, thinking section). I guess I could do that with a regex but it would be better if it was an option in the groq integration module.