Verdugie/Opus-Candid-MoE-V3
LoRA-style vs Qwen/Qwen3-30B-A3B. Verdugie/Opus-Candid-MoE-V3 is a GGUF (Q8_0) of different weights. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of k_proj, o_proj, q_proj, v_proj changed, in 48 of 48 layers; norms, biases and embeddings are untouched.
Tensors changed: Norms 0/193, MLP 0/48, Attention 192/192