6 Downloads Updated yesterday
ollama run reecdev/llama3.2-unmasked
ollama launch claude --model reecdev/llama3.2-unmasked
ollama launch openclaw --model reecdev/llama3.2-unmasked
ollama launch hermes --model reecdev/llama3.2-unmasked
ollama launch codex --model reecdev/llama3.2-unmasked
ollama launch opencode --model reecdev/llama3.2-unmasked
An “unmasked” version of Llama3.2.
Unmasking is a finetuning method similar to abliteration/jailbreaking that removes the original RLHF completely from a model instead of just removing refusals and then doing another finetuning step to make it conversational again.