129 likesNew · 3d old
AtomicChat releases uncensored 1.7B LLM, drops safety refusals
What is LFM3-1.7B-Uncensored-GGUF?
The model is a 1.7 billion-parameter language model with all safety filters stripped, so it will answer any prompt, even harmful ones, while keeping its MMLU score. It’s distributed as GGUF quantized files that run with llama.cpp.
Who it is for
Researchers or hobbyists who need an unrestricted offline model can experiment with it, but must handle the output responsibly.