prithivMLmods/Gemma4-BLIP3o-Captioner-8B
Abliterated (fewer layers) vs google/gemma-4-E4B-it. prithivMLmods/Gemma4-BLIP3o-Captioner-8B matches part of google/gemma-4-E4B-it (677 of its 2130 tensors; the rest of google/gemma-4-E4B-it has no counterpart). On the shared part: prithivMLmods/Gemma4-BLIP3o-Captioner-8B is a GGUF (F16) of different weights. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 35 of 42 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/2, Norms 0/296, MLP 35/126, other 0/85, Attention 35/168