DPO recipe saves a float32 model

Hello,

I have been using the Zephry DPO recipe and the models I get are save in float32. I am using config_full and accelerate multi_gpu.yaml

I think the issue is that the config_full has [not setup the model as bfloat](https://github.com/huggingface/alignment-handbook/blob/87cc800498b17432cfb7f5acb5e9a79f15c867fc/recipes/zephyr-7b-beta/dpo/config_full.yaml#L3)

Should this be changed?