Chompa1422/Qwen3.5-122B-A10B-abliterated
LoRA-style vs Qwen/Qwen3.5-122B-A10B. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of in_proj_qkv, o_proj, out_proj, q_proj changed, in 5 of 48 layers; norms, biases and embeddings are untouched.
Tensors changed: MLP 0/1221, lm_head 0/1, Embeddings 0/5, Attention 10/448, other 0/1, Norms 0/273
5 derivatives checked
- mradermacher/Qwen3.5-122B-A10B-abliterated-i1-GGUF: Quantization
- mradermacher/Qwen3.5-122B-A10B-abliterated-GGUF: Quantization
- TNT3530/Qwen3.5-122B-A10B-abliterated-GGUF: Quantization
- Shifusen/Qwen3.5-122B-A10B-abliterated-FP8: Quantization
- lemuralabs/Qwen3.5-122B-A10B-Abliterated-MLX-3: Quantization
1 other repos hold byte-for-byte identical weight files.