mlabonne/gemma-3-4b-it-abliterated-GGUF
Abliterated (fewer layers) vs google/gemma-3-4b-it. mlabonne/gemma-3-4b-it-abliterated-GGUF matches part of google/gemma-3-4b-it (444 of its 883 tensors; the rest of google/gemma-3-4b-it has no counterpart). On the shared part: mlabonne/gemma-3-4b-it-abliterated-GGUF is a GGUF (FP16) of different weights. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 21 of 34 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/1, Norms 0/205, MLP 21/102, Attention 21/136