Difference between these models and https://huggingface.co/NousResearch/Nous-Hermes-Llama2-13b-GGML?
#2
by
cargo1995
- opened
What's the difference between say, the q4_k_m model in this repo and the original one?
They released their own GGML, but only in one format. Then people asked me if I could do it too as they only had one, so I did my full set.
I would have assumed my q4_K_M would be 100% identical to theirs, but they have different sha256sums. But they're both made from the same model files, so that might just be that we used a slightly different version of llama.cpp that perhaps caused some small difference in the header or something like that.
I don't expect there to be any functional difference between them, so if you already have theirs, there's no need to download mine, and vice versa.
Ok, thanks for the info! And good work!