tirthfx/gemma-3-4b-it-ablated-GGUF
Abliterated (fewer layers) vs google/gemma-3-4b-it. tirthfx/gemma-3-4b-it-ablated-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: tirthfx/gemma-3-4b-it-ablated-GGUF is a GGUF (F16/F32) of different weights. Looks like abliteration (refusal removal): only the output projections (down_proj, embed_tokens, o_proj) changed, in 34 of 34 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 1/1, Norms 0/205, MLP 34/102, Attention 34/136