shaikhsalman/ZabaanAI-Urdu-3B
Adapter vs Qwen/Qwen2.5-3B-Instruct. shaikhsalman/ZabaanAI-Urdu-3B is a LoRA adapter, not full weights: rank 16, on down_proj, gate_proj, k_proj, o_proj, q_proj, up_proj, v_proj in 36 layers (252 adapted matrices). It modifies Qwen/Qwen2.5-3B-Instruct at load time.