AIが動く。未来が更新される。gemma-3-27b-it-abliterated-GGUF
This is an uncensored version of google/gemma-3-27b-it created with a new abliteration technique. See this article to know more about abliteration. I was playing with model weights and noticed that Gemma 3 was much more resilient to abliteration than other models like Qwen 2.5. I experimented with a few recipes to remove refusals while preserving most of the model capabilities. In the original technique, a refusal direction is computed by comparing the residual streams between target (harmful) and baseline (harmless) samples.
翻訳準備中のため原文を表示しています
This is an uncensored version of google/gemma-3-27b-it created with a new abliteration technique. See this article to know more about abliteration. I was playing with model weights and noticed that Gemma 3 was much more resilient to abliteration than other models like Qwen 2.5. I experimented with a few recipes to remove refusals while preserving most of the model capabilities. In the original technique, a refusal direction is computed by comparing the residual streams between target (harmful) and baseline (harmless) samples.