MoA: add the ability to combine multiple LLMs mixture of agents!

[Am0stafa]

With general models like gpt4 we can combine multiple of them to solve a problem. Here is a paper talking about that:

https://arxiv.org/abs/2406.04692