YixuanEvenXu/Qwen3-4B-Base-HIP-adapter
Adapter vs Qwen/Qwen3-4B-Base. YixuanEvenXu/Qwen3-4B-Base-HIP-adapter is a LoRA adapter, not full weights: rank 128, on down_proj, gate_proj, k_proj, o_proj, q_proj, up_proj, v_proj in 36 layers (252 adapted matrices). It modifies Qwen/Qwen3-4B-Base at load time.