prithivMLmods/ultragemma4-e2b-heretic-uncensored
Abliterated (fewer layers) vs google/gemma-4-E2B-it. prithivMLmods/ultragemma4-e2b-heretic-uncensored matches part of google/gemma-4-E2B-it (505 of its 2011 tensors; the rest of google/gemma-4-E2B-it has no counterpart). On the shared part: prithivMLmods/ultragemma4-e2b-heretic-uncensored is a GGUF (F16) of different weights. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj) changed, in 25 of 35 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/2, Norms 0/227, MLP 25/105, other 0/71, Attention 25/100