huihui-ai/Huihui-Ornith-1.5-35B-A3B-abliterated
Abliterated vs ornith-ai/Ornith-1.5-35B-A3B. Looks like abliteration (refusal removal): only the output projections (down_proj, o_proj, out_proj) changed, in 18 of 40 layers. Abliteration edits exactly these matrices to remove a "refusal direction"; everything else is untouched.
Tensors changed: Embeddings 0/5, Norms 0/247, Attention 18/392, MLP 36/1165, lm_head 0/1, other 0/1
19 derivatives checked
- peculiar-ragdoll/Cyber-Tiel-Coder-35B-A3B-GGUF-MTP: Quantization
- peculiar-ragdoll/Cyber-Tiel-Coder-35B-A3B-GGUF: Quantization
- gbuzhf/Ornith-1.5-35B-A3B-Abliterated-CyberTiel-Calibrated-MTPv2-ICE-GGUF: Same weights, converted
- mradermacher/Huihui-Ornith-1.5-35B-A3B-abliterated-GGUF: Quantization
- ArtomYuan/Ornith-1.5-35B-A3B-abliterated-ROCmFPX: Same weights, converted
- mradermacher/Huihui-Ornith-1.5-35B-A3B-abliterated-i1-GGUF: Quantization
- peculiar-ragdoll/Cyber-Tiel-Coder-35B-A3B-MLX-oQ4e-MTP: Quantization
- peculiar-ragdoll/Cyber-Tiel-Coder-35B-A3B-MLX-oQ6e-MTP: Quantization
- peculiar-ragdoll/Cyber-Tiel-Coder-35B-A3B-MLX-oQ4e: Quantization
- r3lax/Huihui-Ornith-1.5-35B-A3B-abliterated-MTPLX-mixed8-4bit-FP16: Quantization
- peculiar-ragdoll/Cyber-Tiel-Coder-35B-A3B-MLX-oQ6e: Quantization
- nuoram/Ornith-1.5-35B-A3B-abliterated-W8A8: Quantization
- 3MPER0RR/Ornith-1.5-35B-A3B-3MPER0RR-abliterated: Abliterated (fewer layers)
- funnygeeker/Huihui-Ornith-1.5-35B-A3B-abliterated-oQ5e-fp16-mtp: Quantization
- jhone888/Ornith-1.5-35B-A3B-FULL-OBLITERATED-NVFP4: Quantization
- Shiftedx/Cyber-Tiel-Coder-35B-A3B-MLX-Attention8-BF16Recurrence-MTP: Quantization
- jhone888/Ornith-1.5-35B-A3B-FULLY-OBLITERATED: Partial fine-tune
- ArkhAngelLifeJiggy/Cyber-Tiel-Coder-35B-A3B-GGUF: Quantization
- dopaemon/Cyber-Tiel-Coder-35B-A3B-MLX-oQ6e-MTP: Quantization
1 other repos hold byte-for-byte identical weight files.