prithivMLmods/Qwen3.6-35B-A3B-Uncensored-Aggressive
Abliterated (fewer layers) vs Qwen/Qwen3.6-35B-A3B. Qwen/Qwen3.6-35B-A3B matches part of prithivMLmods/Qwen3.6-35B-A3B-Uncensored-Aggressive (946 of its 31666 tensors; the rest of prithivMLmods/Qwen3.6-35B-A3B-Uncensored-Aggressive has no counterpart). On the shared part: Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj, out_proj) changed, in 40 of 40 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/4, Attention 40/388, MLP 40/312, Norms 0/241, lm_head 0/1
4 derivatives checked
- mradermacher/Qwen3.6-35B-A3B-Uncensored-Aggressive-i1-GGUF: Quantization
- mradermacher/Qwen3.6-35B-A3B-Uncensored-Aggressive-GGUF: Quantization
- prithivMLmods/Qwen3.6-35B-A3B-Uncensored-Aggressive-GGUF: Quantization
- prithivMLmods/Q3.6-35B-A3B-abliterated-0520-MAX-STOR-check: Same weights, repacked (fewer layers)