Already exists
#2
by
lunahr
- opened
I was first to abliterate this model: https://huggingface.co/lunahr/Hermes-3-Llama-3.2-3B-abliterated
We support your contribution to the community, and we are providing it for testing purposes only.
yours may have a different behavior as checksums differ so the weights aren't a 1:1 recreation
that may be cuz abliteration is not reproducible because of LLM answers being different every time you infer on it
yes