{ "architectures": [ "MageVLForConditionalGeneration" ], "bos_token_id": 151643, "dtype": "bfloat16", "eos_token_id": 151645, "image_token_id": 151655, "model_type": "mage_vl", "quantization": { "group_size": 64, "bits": 4, "mode": "affine", "model.embed_tokens": { "bits": 8, "group_size": 64 }, "model.layers.0.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.0.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.0.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.0.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.0.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.0.mlp.down_proj": { "bits": 8, "group_size": 64 }, "model.layers.0.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.1.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.1.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.1.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.1.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.1.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.1.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.1.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.2.self_attn.q_proj": { "bits": 8, "group_size": 64 }, "model.layers.2.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.2.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.2.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.2.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.2.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.2.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.3.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.3.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.3.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.3.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.3.mlp.gate_proj": { "bits": 8, "group_size": 64 }, "model.layers.3.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.3.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.4.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.4.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.4.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.4.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.4.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.4.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.4.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.5.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.5.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.5.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.5.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.5.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.5.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.5.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.6.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.6.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.6.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.6.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.6.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.6.mlp.down_proj": { "bits": 8, "group_size": 64 }, "model.layers.6.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.7.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.7.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.7.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.7.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.7.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.7.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.7.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.8.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.8.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.8.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.8.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.8.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.8.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.8.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.9.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.9.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.9.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.9.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.9.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.9.mlp.down_proj": { "bits": 8, "group_size": 64 }, "model.layers.9.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.10.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.10.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.10.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.10.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.10.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.10.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.10.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.11.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.11.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.11.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.11.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.11.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.11.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.11.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.12.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.12.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.12.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.12.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.12.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.12.mlp.down_proj": { "bits": 8, "group_size": 64 }, "model.layers.12.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.13.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.13.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.13.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.13.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.13.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.13.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.13.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.14.self_attn.q_proj": { "bits": 8, "group_size": 64 }, "model.layers.14.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.14.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.14.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.14.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.14.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.14.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.15.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.15.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.15.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.15.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.15.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.15.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.15.mlp.up_proj": { "bits": 8, "group_size": 64 }, "model.layers.16.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.16.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.16.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.16.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.16.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.16.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.16.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.17.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.17.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.17.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.17.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.17.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.17.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.17.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.18.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.18.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.18.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.18.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.18.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.18.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.18.mlp.up_proj": { "bits": 8, "group_size": 64 }, "model.layers.19.self_attn.q_proj": { "bits": 8, "group_size": 64 }, "model.layers.19.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.19.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.19.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.19.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.19.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.19.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.20.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.20.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.20.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.20.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.20.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.20.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.20.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.21.self_attn.q_proj": { "bits": 8, "group_size": 64 }, "model.layers.21.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.21.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.21.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.21.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.21.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.21.mlp.up_proj": { "bits": 8, "group_size": 64 }, "model.layers.22.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.22.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.22.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.22.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.22.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.22.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.22.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.23.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.23.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.23.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.23.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.23.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.23.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.23.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.24.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.24.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.24.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.24.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.24.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.24.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.24.mlp.up_proj": { "bits": 8, "group_size": 64 }, "model.layers.25.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.25.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.25.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.25.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.25.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.25.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.25.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.26.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.26.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.26.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.26.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.26.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.26.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.26.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.27.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.27.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.27.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.27.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.27.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.27.mlp.down_proj": { "bits": 8, "group_size": 64 }, "model.layers.27.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.28.self_attn.q_proj": { "bits": 8, "group_size": 64 }, "model.layers.28.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.28.self_attn.v_proj": { "bits": 4, "group_size": 64 }, "model.layers.28.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.28.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.28.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.28.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.29.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.29.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.29.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.29.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.29.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.29.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.29.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.30.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.30.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.30.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.30.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.30.mlp.gate_proj": { "bits": 8, "group_size": 64 }, "model.layers.30.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.30.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.31.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.31.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.31.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.31.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.31.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.31.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.31.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.32.self_attn.q_proj": { "bits": 8, "group_size": 64 }, "model.layers.32.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.32.self_attn.v_proj": { "bits": 4, "group_size": 64 }, "model.layers.32.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.32.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.32.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.32.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.33.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.33.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.33.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.33.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.33.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.33.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.33.mlp.up_proj": { "bits": 8, "group_size": 64 }, "model.layers.34.self_attn.q_proj": { "bits": 8, "group_size": 64 }, "model.layers.34.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.34.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.34.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.34.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.34.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.34.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.35.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.35.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.35.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.35.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.35.mlp.gate_proj": { "bits": 8, "group_size": 64 }, "model.layers.35.mlp.down_proj": { "bits": 8, "group_size": 64 }, "model.layers.35.mlp.up_proj": { "bits": 8, "group_size": 64 }, "lm_head": { "bits": 8, "group_size": 64 } }, "quantization_config": { "group_size": 64, "bits": 4, "mode": "affine", "model.embed_tokens": { "bits": 8, "group_size": 64 }, "model.layers.0.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.0.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.0.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.0.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.0.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.0.mlp.down_proj": { "bits": 8, "group_size": 64 }, "model.layers.0.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.1.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.1.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.1.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.1.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.1.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.1.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.1.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.2.self_attn.q_proj": { "bits": 8, "group_size": 64 }, "model.layers.2.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.2.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.2.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.2.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.2.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.2.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.3.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.3.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.3.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.3.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.3.mlp.gate_proj": { "bits": 8, "group_size": 64 }, "model.layers.3.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.3.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.4.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.4.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.4.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.4.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.4.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.4.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.4.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.5.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.5.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.5.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.5.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.5.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.5.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.5.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.6.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.6.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.6.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.6.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.6.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.6.mlp.down_proj": { "bits": 8, "group_size": 64 }, "model.layers.6.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.7.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.7.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.7.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.7.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.7.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.7.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.7.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.8.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.8.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.8.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.8.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.8.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.8.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.8.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.9.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.9.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.9.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.9.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.9.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.9.mlp.down_proj": { "bits": 8, "group_size": 64 }, "model.layers.9.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.10.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.10.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.10.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.10.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.10.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.10.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.10.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.11.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.11.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.11.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.11.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.11.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.11.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.11.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.12.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.12.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.12.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.12.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.12.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.12.mlp.down_proj": { "bits": 8, "group_size": 64 }, "model.layers.12.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.13.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.13.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.13.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.13.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.13.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.13.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.13.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.14.self_attn.q_proj": { "bits": 8, "group_size": 64 }, "model.layers.14.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.14.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.14.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.14.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.14.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.14.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.15.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.15.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.15.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.15.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.15.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.15.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.15.mlp.up_proj": { "bits": 8, "group_size": 64 }, "model.layers.16.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.16.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.16.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.16.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.16.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.16.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.16.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.17.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.17.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.17.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.17.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.17.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.17.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.17.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.18.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.18.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.18.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.18.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.18.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.18.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.18.mlp.up_proj": { "bits": 8, "group_size": 64 }, "model.layers.19.self_attn.q_proj": { "bits": 8, "group_size": 64 }, "model.layers.19.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.19.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.19.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.19.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.19.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.19.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.20.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.20.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.20.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.20.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.20.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.20.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.20.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.21.self_attn.q_proj": { "bits": 8, "group_size": 64 }, "model.layers.21.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.21.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.21.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.21.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.21.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.21.mlp.up_proj": { "bits": 8, "group_size": 64 }, "model.layers.22.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.22.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.22.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.22.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.22.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.22.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.22.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.23.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.23.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.23.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.23.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.23.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.23.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.23.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.24.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.24.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.24.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.24.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.24.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.24.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.24.mlp.up_proj": { "bits": 8, "group_size": 64 }, "model.layers.25.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.25.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.25.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.25.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.25.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.25.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.25.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.26.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.26.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.26.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.26.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.26.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.26.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.26.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.27.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.27.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.27.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.27.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.27.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.27.mlp.down_proj": { "bits": 8, "group_size": 64 }, "model.layers.27.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.28.self_attn.q_proj": { "bits": 8, "group_size": 64 }, "model.layers.28.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.28.self_attn.v_proj": { "bits": 4, "group_size": 64 }, "model.layers.28.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.28.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.28.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.28.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.29.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.29.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.29.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.29.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.29.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.29.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.29.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.30.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.30.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.30.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.30.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.30.mlp.gate_proj": { "bits": 8, "group_size": 64 }, "model.layers.30.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.30.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.31.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.31.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.31.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.31.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.31.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.31.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.31.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.32.self_attn.q_proj": { "bits": 8, "group_size": 64 }, "model.layers.32.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.32.self_attn.v_proj": { "bits": 4, "group_size": 64 }, "model.layers.32.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.32.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.32.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.32.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.33.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.33.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.33.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.33.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.33.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.33.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.33.mlp.up_proj": { "bits": 8, "group_size": 64 }, "model.layers.34.self_attn.q_proj": { "bits": 8, "group_size": 64 }, "model.layers.34.self_attn.k_proj": { "bits": 4, "group_size": 64 }, "model.layers.34.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.34.self_attn.o_proj": { "bits": 4, "group_size": 64 }, "model.layers.34.mlp.gate_proj": { "bits": 4, "group_size": 64 }, "model.layers.34.mlp.down_proj": { "bits": 4, "group_size": 64 }, "model.layers.34.mlp.up_proj": { "bits": 4, "group_size": 64 }, "model.layers.35.self_attn.q_proj": { "bits": 4, "group_size": 64 }, "model.layers.35.self_attn.k_proj": { "bits": 8, "group_size": 64 }, "model.layers.35.self_attn.v_proj": { "bits": 8, "group_size": 64 }, "model.layers.35.self_attn.o_proj": { "bits": 8, "group_size": 64 }, "model.layers.35.mlp.gate_proj": { "bits": 8, "group_size": 64 }, "model.layers.35.mlp.down_proj": { "bits": 8, "group_size": 64 }, "model.layers.35.mlp.up_proj": { "bits": 8, "group_size": 64 }, "lm_head": { "bits": 8, "group_size": 64 } }, "text_config": { "_name_or_path": "/mage-vl/pretrain_models/Qwen3-4B-Instruct-2507", "architectures": [ "MageVLForConditionalGeneration" ], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 151643, "dtype": "bfloat16", "eos_token_id": 151645, "head_dim": 128, "hidden_act": "silu", "hidden_size": 2560, "initializer_range": 0.02, "intermediate_size": 9728, "layer_types": [ "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention", "full_attention" ], "max_position_embeddings": 262144, "max_window_layers": 36, "model_type": "qwen3", "num_attention_heads": 32, "num_hidden_layers": 36, "num_key_value_heads": 8, "rms_norm_eps": 1e-06, "rope_scaling": null, "sliding_window": null, "use_cache": true, "use_sliding_window": false, "vocab_size": 151936, "rope_parameters": { "rope_theta": 5000000, "rope_type": "default" }, "tie_word_embeddings": false }, "tie_word_embeddings": false, "transformers_version": "5.7.0", "video_token_id": 151656, "vision_end_token_id": 151653, "vision_start_token_id": 151652, "vision_config": { "attention_dropout": 0.0, "frame_windows_size": 4, "hidden_act": "gelu", "hidden_size": 1024, "image_size": 448, "initializer_range": 0.02, "intermediate_size": 4096, "layer_norm_eps": 1e-06, "layer_norm_type": "layer_norm", "model_type": "mage_vl_vision", "num_attention_heads": 16, "num_channels": 3, "num_hidden_layers": 24, "out_hidden_size": 2560, "patch_size": 16, "rope_theta": 10000.0, "spatial_merge_size": 2, "temporal_patch_size": 1, "text_hidden_size": 2560, "tokens_per_second": 1, "use_head": false, "max_position_embeddings": 8192 }, "optiq_vision": { "sidecar": "optiq/optiq_vision.safetensors", "dtype": "bfloat16", "n_tensors": 297, "base_model": "microsoft/Mage-VL" } }