LunarMoonar/Huihui-Qwen3.8-27B-abliterated-GGUF
LoRA-style vs Qwen/Qwen3.8-27B. LunarMoonar/Huihui-Qwen3.8-27B-abliterated-GGUF is a GGUF (Q8_0_L) of different weights. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, k_proj, o_proj, q_proj, up_proj, v_proj changed, in 64 of 64 layers; norms, biases and embeddings are untouched.
Tensors changed: Norms 0/161, MLP 177/192, Attention 61/64, Embeddings 0/1, lm_head 0/1