I am just loading GGUF models from HuggingFace that have good scores in the benchmarks, and running my private eval set from my current project. Some of the me

I am just loading GGUF models from HuggingFace that have good scores in the benchmarks, and running my private eval set from my current project. Some of the merged models are surprisingly good compared with simple fine-tunes.