Text this: Mixture of prompts learning for vision-language models