dx8152/Qwen-Image-Edit-2509-Fusion
Adapter vs Qwen/Qwen-Image-Edit-2509. dx8152/Qwen-Image-Edit-2509-Fusion is a LoRA adapter, not full weights: rank 16, on 0, 1, 2, add_k_proj, add_q_proj, add_v_proj, to_add_out, to_k… (720 adapted matrices). It modifies Qwen/Qwen-Image-Edit-2509 at load time.
2 other repos hold byte-for-byte identical weight files.