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