Fetching the paper…

Model ensemble instead of prompt fusion: a sample-specific knowledge transfer method for few-shot prompt tuning · Around