{ "add_skip_keys": false, "dequantize_fp32": false, "dynamic_loss_threshold": 0.01, "group_size": 0, "is_integer": true, "is_training": false, "modules_dtype_dict": { "float5_e3m1fn": [ "model.layers.0.mlp.experts.9.down_proj.weight", "model.layers.0.mlp.experts.12.down_proj.weight" ], "int5": [ "vision_model.encoder.layers.0.self_attn.k_proj.weight", "vision_model.encoder.layers.0.self_attn.v_proj.weight", "vision_model.encoder.layers.0.self_attn.q_proj.weight", "vision_model.encoder.layers.0.mlp.fc1.weight", "vision_model.encoder.layers.0.mlp.fc2.weight", "vision_model.encoder.layers.1.self_attn.k_proj.weight", "vision_model.encoder.layers.1.self_attn.v_proj.weight", "vision_model.encoder.layers.1.self_attn.q_proj.weight", "vision_model.encoder.layers.1.mlp.fc2.weight", "vision_model.encoder.layers.2.mlp.fc2.weight", "vision_model.encoder.layers.3.mlp.fc2.weight", "vision_model.encoder.layers.4.mlp.fc2.weight", "vision_model.encoder.layers.5.mlp.fc2.weight", "vision_model.encoder.layers.6.mlp.fc2.weight", "vision_model.encoder.layers.7.mlp.fc2.weight", "vision_model.encoder.layers.8.mlp.fc2.weight", "vision_model.encoder.layers.9.mlp.fc2.weight", "vision_model.encoder.layers.10.mlp.fc2.weight", "vision_model.encoder.layers.11.mlp.fc2.weight", "vision_model.encoder.layers.12.mlp.fc2.weight", "vision_model.encoder.layers.13.mlp.fc2.weight", "vision_model.encoder.layers.14.mlp.fc2.weight", "vision_model.encoder.layers.15.mlp.fc2.weight", "vision_model.encoder.layers.16.mlp.fc2.weight", "vision_model.encoder.layers.17.mlp.fc2.weight", "vision_model.encoder.layers.18.mlp.fc2.weight", "vision_model.encoder.layers.19.mlp.fc2.weight", "vision_model.encoder.layers.20.mlp.fc2.weight", "vision_model.encoder.layers.21.mlp.fc2.weight", "vision_model.encoder.layers.22.mlp.fc2.weight", "vision_model.encoder.layers.23.mlp.fc2.weight", "vision_model.encoder.layers.24.mlp.fc2.weight", "vision_model.encoder.layers.25.mlp.fc2.weight", "vision_model.encoder.layers.26.mlp.fc2.weight", "model.layers.0.mlp.gate.wg.weight", "model.layers.0.mlp.experts.2.gate_and_up_proj.weight", "model.layers.0.mlp.experts.2.down_proj.weight", "model.layers.0.mlp.experts.3.down_proj.weight", "model.layers.0.mlp.experts.7.down_proj.weight", "model.layers.0.mlp.experts.9.gate_and_up_proj.weight", "model.layers.0.mlp.experts.12.gate_and_up_proj.weight", "model.layers.0.mlp.experts.14.down_proj.weight", "model.layers.0.mlp.experts.25.gate_and_up_proj.weight", "model.layers.0.mlp.experts.25.down_proj.weight", "model.layers.0.mlp.experts.35.down_proj.weight", "model.layers.0.mlp.experts.36.down_proj.weight", "model.layers.0.mlp.experts.39.down_proj.weight", "model.layers.0.mlp.experts.61.down_proj.weight", "model.layers.1.mlp.gate.wg.weight", "model.layers.2.mlp.gate.wg.weight", "model.layers.3.mlp.gate.wg.weight", "model.layers.4.mlp.gate.wg.weight", "model.layers.5.mlp.gate.wg.weight", "model.layers.6.mlp.gate.wg.weight", "model.layers.7.mlp.gate.wg.weight", "model.layers.8.mlp.gate.wg.weight", "model.layers.9.mlp.gate.wg.weight", "model.layers.10.mlp.gate.wg.weight", "model.layers.11.mlp.gate.wg.weight", "model.layers.12.mlp.gate.wg.weight", "model.layers.13.mlp.gate.wg.weight", "model.layers.14.mlp.gate.wg.weight", "model.layers.15.mlp.gate.wg.weight", "model.layers.16.mlp.gate.wg.weight", "model.layers.17.mlp.gate.wg.weight", "model.layers.18.mlp.gate.wg.weight", "model.layers.19.mlp.gate.wg.weight", "model.layers.20.mlp.gate.wg.weight", "model.layers.21.mlp.gate.wg.weight", "model.layers.22.mlp.gate.wg.weight", "model.layers.23.mlp.gate.wg.weight", "model.layers.24.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.24.mlp.gate.wg.weight", "model.layers.25.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.25.mlp.gate.wg.weight", "model.layers.26.mlp.gate.wg.weight", "model.layers.27.mlp.shared_mlp.down_proj.weight", "model.layers.27.mlp.gate.wg.weight", "model.layers.28.mlp.shared_mlp.down_proj.weight", "model.layers.28.mlp.gate.wg.weight", "model.layers.29.mlp.shared_mlp.down_proj.weight", "model.layers.29.mlp.gate.wg.weight", "model.layers.30.mlp.shared_mlp.down_proj.weight", "model.layers.30.mlp.gate.wg.weight", "model.layers.31.mlp.shared_mlp.down_proj.weight" ], "uint4": [ "vision_model.encoder.layers.0.self_attn.out_proj.weight", "vision_model.encoder.layers.1.self_attn.out_proj.weight", "vision_model.encoder.layers.1.mlp.fc1.weight", "vision_model.encoder.layers.2.self_attn.k_proj.weight", "vision_model.encoder.layers.2.self_attn.v_proj.weight", "vision_model.encoder.layers.2.self_attn.q_proj.weight", "vision_model.encoder.layers.2.self_attn.out_proj.weight", "vision_model.encoder.layers.2.mlp.fc1.weight", "vision_model.encoder.layers.3.self_attn.k_proj.weight", "vision_model.encoder.layers.3.self_attn.v_proj.weight", "vision_model.encoder.layers.3.self_attn.q_proj.weight", "vision_model.encoder.layers.3.self_attn.out_proj.weight", "vision_model.encoder.layers.3.mlp.fc1.weight", "vision_model.encoder.layers.4.self_attn.k_proj.weight", "vision_model.encoder.layers.4.self_attn.v_proj.weight", "vision_model.encoder.layers.4.self_attn.q_proj.weight", "vision_model.encoder.layers.4.self_attn.out_proj.weight", "vision_model.encoder.layers.4.mlp.fc1.weight", "vision_model.encoder.layers.5.self_attn.k_proj.weight", "vision_model.encoder.layers.5.self_attn.v_proj.weight", "vision_model.encoder.layers.5.self_attn.q_proj.weight", "vision_model.encoder.layers.5.self_attn.out_proj.weight", "vision_model.encoder.layers.5.mlp.fc1.weight", "vision_model.encoder.layers.6.self_attn.k_proj.weight", "vision_model.encoder.layers.6.self_attn.v_proj.weight", "vision_model.encoder.layers.6.self_attn.q_proj.weight", "vision_model.encoder.layers.6.self_attn.out_proj.weight", "vision_model.encoder.layers.6.mlp.fc1.weight", "vision_model.encoder.layers.7.self_attn.k_proj.weight", "vision_model.encoder.layers.7.self_attn.v_proj.weight", "vision_model.encoder.layers.7.self_attn.q_proj.weight", "vision_model.encoder.layers.7.self_attn.out_proj.weight", "vision_model.encoder.layers.7.mlp.fc1.weight", "vision_model.encoder.layers.8.self_attn.k_proj.weight", "vision_model.encoder.layers.8.self_attn.v_proj.weight", "vision_model.encoder.layers.8.self_attn.q_proj.weight", "vision_model.encoder.layers.8.self_attn.out_proj.weight", "vision_model.encoder.layers.8.mlp.fc1.weight", "vision_model.encoder.layers.9.self_attn.k_proj.weight", "vision_model.encoder.layers.9.self_attn.v_proj.weight", "vision_model.encoder.layers.9.self_attn.q_proj.weight", "vision_model.encoder.layers.9.self_attn.out_proj.weight", "vision_model.encoder.layers.9.mlp.fc1.weight", "vision_model.encoder.layers.10.self_attn.k_proj.weight", "vision_model.encoder.layers.10.self_attn.v_proj.weight", "vision_model.encoder.layers.10.self_attn.q_proj.weight", "vision_model.encoder.layers.10.self_attn.out_proj.weight", "vision_model.encoder.layers.10.mlp.fc1.weight", "vision_model.encoder.layers.11.self_attn.k_proj.weight", "vision_model.encoder.layers.11.self_attn.v_proj.weight", "vision_model.encoder.layers.11.self_attn.q_proj.weight", "vision_model.encoder.layers.11.self_attn.out_proj.weight", "vision_model.encoder.layers.11.mlp.fc1.weight", "vision_model.encoder.layers.12.self_attn.k_proj.weight", "vision_model.encoder.layers.12.self_attn.v_proj.weight", "vision_model.encoder.layers.12.self_attn.q_proj.weight", "vision_model.encoder.layers.12.self_attn.out_proj.weight", "vision_model.encoder.layers.12.mlp.fc1.weight", "vision_model.encoder.layers.13.self_attn.k_proj.weight", "vision_model.encoder.layers.13.self_attn.v_proj.weight", "vision_model.encoder.layers.13.self_attn.q_proj.weight", "vision_model.encoder.layers.13.self_attn.out_proj.weight", "vision_model.encoder.layers.13.mlp.fc1.weight", "vision_model.encoder.layers.14.self_attn.k_proj.weight", "vision_model.encoder.layers.14.self_attn.v_proj.weight", "vision_model.encoder.layers.14.self_attn.q_proj.weight", "vision_model.encoder.layers.14.self_attn.out_proj.weight", "vision_model.encoder.layers.14.mlp.fc1.weight", "vision_model.encoder.layers.15.self_attn.k_proj.weight", "vision_model.encoder.layers.15.self_attn.v_proj.weight", "vision_model.encoder.layers.15.self_attn.q_proj.weight", "vision_model.encoder.layers.15.self_attn.out_proj.weight", "vision_model.encoder.layers.15.mlp.fc1.weight", "vision_model.encoder.layers.16.self_attn.k_proj.weight", "vision_model.encoder.layers.16.self_attn.v_proj.weight", "vision_model.encoder.layers.16.self_attn.q_proj.weight", "vision_model.encoder.layers.16.self_attn.out_proj.weight", "vision_model.encoder.layers.16.mlp.fc1.weight", "vision_model.encoder.layers.17.self_attn.k_proj.weight", "vision_model.encoder.layers.17.self_attn.v_proj.weight", "vision_model.encoder.layers.17.self_attn.q_proj.weight", "vision_model.encoder.layers.17.self_attn.out_proj.weight", "vision_model.encoder.layers.17.mlp.fc1.weight", "vision_model.encoder.layers.18.self_attn.k_proj.weight", "vision_model.encoder.layers.18.self_attn.v_proj.weight", "vision_model.encoder.layers.18.self_attn.q_proj.weight", "vision_model.encoder.layers.18.self_attn.out_proj.weight", "vision_model.encoder.layers.18.mlp.fc1.weight", "vision_model.encoder.layers.19.self_attn.k_proj.weight", "vision_model.encoder.layers.19.self_attn.v_proj.weight", "vision_model.encoder.layers.19.self_attn.q_proj.weight", "vision_model.encoder.layers.19.self_attn.out_proj.weight", "vision_model.encoder.layers.19.mlp.fc1.weight", "vision_model.encoder.layers.20.self_attn.k_proj.weight", "vision_model.encoder.layers.20.self_attn.v_proj.weight", "vision_model.encoder.layers.20.self_attn.q_proj.weight", "vision_model.encoder.layers.20.self_attn.out_proj.weight", "vision_model.encoder.layers.20.mlp.fc1.weight", "vision_model.encoder.layers.21.self_attn.k_proj.weight", "vision_model.encoder.layers.21.self_attn.v_proj.weight", "vision_model.encoder.layers.21.self_attn.q_proj.weight", "vision_model.encoder.layers.21.self_attn.out_proj.weight", "vision_model.encoder.layers.21.mlp.fc1.weight", "vision_model.encoder.layers.22.self_attn.k_proj.weight", "vision_model.encoder.layers.22.self_attn.v_proj.weight", "vision_model.encoder.layers.22.self_attn.q_proj.weight", "vision_model.encoder.layers.22.self_attn.out_proj.weight", "vision_model.encoder.layers.22.mlp.fc1.weight", "vision_model.encoder.layers.23.self_attn.k_proj.weight", "vision_model.encoder.layers.23.self_attn.v_proj.weight", "vision_model.encoder.layers.23.self_attn.q_proj.weight", "vision_model.encoder.layers.23.self_attn.out_proj.weight", "vision_model.encoder.layers.23.mlp.fc1.weight", "vision_model.encoder.layers.24.self_attn.k_proj.weight", "vision_model.encoder.layers.24.self_attn.v_proj.weight", "vision_model.encoder.layers.24.self_attn.q_proj.weight", "vision_model.encoder.layers.24.self_attn.out_proj.weight", "vision_model.encoder.layers.24.mlp.fc1.weight", "vision_model.encoder.layers.25.self_attn.k_proj.weight", "vision_model.encoder.layers.25.self_attn.v_proj.weight", "vision_model.encoder.layers.25.self_attn.q_proj.weight", "vision_model.encoder.layers.25.self_attn.out_proj.weight", "vision_model.encoder.layers.25.mlp.fc1.weight", "vision_model.encoder.layers.26.self_attn.k_proj.weight", "vision_model.encoder.layers.26.self_attn.v_proj.weight", "vision_model.encoder.layers.26.self_attn.q_proj.weight", "vision_model.encoder.layers.26.self_attn.out_proj.weight", "vision_model.encoder.layers.26.mlp.fc1.weight", "model.layers.0.self_attn.qkv_proj.weight", "model.layers.0.self_attn.o_proj.weight", "model.layers.0.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.0.mlp.shared_mlp.down_proj.weight", "model.layers.0.mlp.experts.0.gate_and_up_proj.weight", "model.layers.0.mlp.experts.0.down_proj.weight", "model.layers.0.mlp.experts.1.gate_and_up_proj.weight", "model.layers.0.mlp.experts.1.down_proj.weight", "model.layers.0.mlp.experts.3.gate_and_up_proj.weight", "model.layers.0.mlp.experts.4.gate_and_up_proj.weight", "model.layers.0.mlp.experts.4.down_proj.weight", "model.layers.0.mlp.experts.5.gate_and_up_proj.weight", "model.layers.0.mlp.experts.5.down_proj.weight", "model.layers.0.mlp.experts.6.gate_and_up_proj.weight", "model.layers.0.mlp.experts.6.down_proj.weight", "model.layers.0.mlp.experts.7.gate_and_up_proj.weight", "model.layers.0.mlp.experts.8.gate_and_up_proj.weight", "model.layers.0.mlp.experts.8.down_proj.weight", "model.layers.0.mlp.experts.10.gate_and_up_proj.weight", "model.layers.0.mlp.experts.10.down_proj.weight", "model.layers.0.mlp.experts.11.gate_and_up_proj.weight", "model.layers.0.mlp.experts.11.down_proj.weight", "model.layers.0.mlp.experts.13.gate_and_up_proj.weight", "model.layers.0.mlp.experts.13.down_proj.weight", "model.layers.0.mlp.experts.14.gate_and_up_proj.weight", "model.layers.0.mlp.experts.15.gate_and_up_proj.weight", "model.layers.0.mlp.experts.15.down_proj.weight", "model.layers.0.mlp.experts.16.gate_and_up_proj.weight", "model.layers.0.mlp.experts.16.down_proj.weight", "model.layers.0.mlp.experts.17.gate_and_up_proj.weight", "model.layers.0.mlp.experts.17.down_proj.weight", "model.layers.0.mlp.experts.18.gate_and_up_proj.weight", "model.layers.0.mlp.experts.18.down_proj.weight", "model.layers.0.mlp.experts.19.gate_and_up_proj.weight", "model.layers.0.mlp.experts.19.down_proj.weight", "model.layers.0.mlp.experts.20.gate_and_up_proj.weight", "model.layers.0.mlp.experts.20.down_proj.weight", "model.layers.0.mlp.experts.21.gate_and_up_proj.weight", "model.layers.0.mlp.experts.21.down_proj.weight", "model.layers.0.mlp.experts.22.gate_and_up_proj.weight", "model.layers.0.mlp.experts.22.down_proj.weight", "model.layers.0.mlp.experts.23.gate_and_up_proj.weight", "model.layers.0.mlp.experts.23.down_proj.weight", "model.layers.0.mlp.experts.24.gate_and_up_proj.weight", "model.layers.0.mlp.experts.24.down_proj.weight", "model.layers.0.mlp.experts.26.gate_and_up_proj.weight", "model.layers.0.mlp.experts.26.down_proj.weight", "model.layers.0.mlp.experts.27.gate_and_up_proj.weight", "model.layers.0.mlp.experts.27.down_proj.weight", "model.layers.0.mlp.experts.28.gate_and_up_proj.weight", "model.layers.0.mlp.experts.28.down_proj.weight", "model.layers.0.mlp.experts.29.gate_and_up_proj.weight", "model.layers.0.mlp.experts.29.down_proj.weight", "model.layers.0.mlp.experts.30.gate_and_up_proj.weight", "model.layers.0.mlp.experts.30.down_proj.weight", "model.layers.0.mlp.experts.31.gate_and_up_proj.weight", "model.layers.0.mlp.experts.31.down_proj.weight", "model.layers.0.mlp.experts.32.gate_and_up_proj.weight", "model.layers.0.mlp.experts.32.down_proj.weight", "model.layers.0.mlp.experts.33.gate_and_up_proj.weight", "model.layers.0.mlp.experts.33.down_proj.weight", "model.layers.0.mlp.experts.34.gate_and_up_proj.weight", "model.layers.0.mlp.experts.34.down_proj.weight", "model.layers.0.mlp.experts.35.gate_and_up_proj.weight", "model.layers.0.mlp.experts.36.gate_and_up_proj.weight", "model.layers.0.mlp.experts.37.gate_and_up_proj.weight", "model.layers.0.mlp.experts.37.down_proj.weight", "model.layers.0.mlp.experts.38.gate_and_up_proj.weight", "model.layers.0.mlp.experts.38.down_proj.weight", "model.layers.0.mlp.experts.39.gate_and_up_proj.weight", "model.layers.0.mlp.experts.40.gate_and_up_proj.weight", "model.layers.0.mlp.experts.40.down_proj.weight", "model.layers.0.mlp.experts.41.gate_and_up_proj.weight", "model.layers.0.mlp.experts.41.down_proj.weight", "model.layers.0.mlp.experts.42.gate_and_up_proj.weight", "model.layers.0.mlp.experts.42.down_proj.weight", "model.layers.0.mlp.experts.43.gate_and_up_proj.weight", "model.layers.0.mlp.experts.43.down_proj.weight", "model.layers.0.mlp.experts.44.gate_and_up_proj.weight", "model.layers.0.mlp.experts.44.down_proj.weight", "model.layers.0.mlp.experts.45.gate_and_up_proj.weight", "model.layers.0.mlp.experts.45.down_proj.weight", "model.layers.0.mlp.experts.46.gate_and_up_proj.weight", "model.layers.0.mlp.experts.46.down_proj.weight", "model.layers.0.mlp.experts.47.gate_and_up_proj.weight", "model.layers.0.mlp.experts.47.down_proj.weight", "model.layers.0.mlp.experts.48.gate_and_up_proj.weight", "model.layers.0.mlp.experts.48.down_proj.weight", "model.layers.0.mlp.experts.49.gate_and_up_proj.weight", "model.layers.0.mlp.experts.49.down_proj.weight", "model.layers.0.mlp.experts.50.gate_and_up_proj.weight", "model.layers.0.mlp.experts.50.down_proj.weight", "model.layers.0.mlp.experts.51.gate_and_up_proj.weight", "model.layers.0.mlp.experts.51.down_proj.weight", "model.layers.0.mlp.experts.52.gate_and_up_proj.weight", "model.layers.0.mlp.experts.52.down_proj.weight", "model.layers.0.mlp.experts.53.gate_and_up_proj.weight", "model.layers.0.mlp.experts.53.down_proj.weight", "model.layers.0.mlp.experts.54.gate_and_up_proj.weight", "model.layers.0.mlp.experts.54.down_proj.weight", "model.layers.0.mlp.experts.55.gate_and_up_proj.weight", "model.layers.0.mlp.experts.55.down_proj.weight", "model.layers.0.mlp.experts.56.gate_and_up_proj.weight", "model.layers.0.mlp.experts.56.down_proj.weight", "model.layers.0.mlp.experts.57.gate_and_up_proj.weight", "model.layers.0.mlp.experts.57.down_proj.weight", "model.layers.0.mlp.experts.58.gate_and_up_proj.weight", "model.layers.0.mlp.experts.58.down_proj.weight", "model.layers.0.mlp.experts.59.gate_and_up_proj.weight", "model.layers.0.mlp.experts.59.down_proj.weight", "model.layers.0.mlp.experts.60.gate_and_up_proj.weight", "model.layers.0.mlp.experts.60.down_proj.weight", "model.layers.0.mlp.experts.61.gate_and_up_proj.weight", "model.layers.0.mlp.experts.62.gate_and_up_proj.weight", "model.layers.0.mlp.experts.62.down_proj.weight", "model.layers.0.mlp.experts.63.gate_and_up_proj.weight", "model.layers.0.mlp.experts.63.down_proj.weight", "model.layers.1.self_attn.qkv_proj.weight", "model.layers.1.self_attn.o_proj.weight", "model.layers.1.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.1.mlp.shared_mlp.down_proj.weight", "model.layers.1.mlp.experts.0.gate_and_up_proj.weight", "model.layers.1.mlp.experts.0.down_proj.weight", "model.layers.1.mlp.experts.1.gate_and_up_proj.weight", "model.layers.1.mlp.experts.1.down_proj.weight", "model.layers.1.mlp.experts.2.gate_and_up_proj.weight", "model.layers.1.mlp.experts.2.down_proj.weight", "model.layers.1.mlp.experts.3.gate_and_up_proj.weight", "model.layers.1.mlp.experts.3.down_proj.weight", "model.layers.1.mlp.experts.4.gate_and_up_proj.weight", "model.layers.1.mlp.experts.4.down_proj.weight", "model.layers.1.mlp.experts.5.gate_and_up_proj.weight", "model.layers.1.mlp.experts.5.down_proj.weight", "model.layers.1.mlp.experts.6.gate_and_up_proj.weight", "model.layers.1.mlp.experts.6.down_proj.weight", "model.layers.1.mlp.experts.7.gate_and_up_proj.weight", "model.layers.1.mlp.experts.7.down_proj.weight", "model.layers.1.mlp.experts.8.gate_and_up_proj.weight", "model.layers.1.mlp.experts.8.down_proj.weight", "model.layers.1.mlp.experts.9.gate_and_up_proj.weight", "model.layers.1.mlp.experts.9.down_proj.weight", "model.layers.1.mlp.experts.10.gate_and_up_proj.weight", "model.layers.1.mlp.experts.10.down_proj.weight", "model.layers.1.mlp.experts.11.gate_and_up_proj.weight", "model.layers.1.mlp.experts.11.down_proj.weight", "model.layers.1.mlp.experts.12.gate_and_up_proj.weight", "model.layers.1.mlp.experts.12.down_proj.weight", "model.layers.1.mlp.experts.13.gate_and_up_proj.weight", "model.layers.1.mlp.experts.13.down_proj.weight", "model.layers.1.mlp.experts.14.gate_and_up_proj.weight", "model.layers.1.mlp.experts.14.down_proj.weight", "model.layers.1.mlp.experts.15.gate_and_up_proj.weight", "model.layers.1.mlp.experts.15.down_proj.weight", "model.layers.1.mlp.experts.16.gate_and_up_proj.weight", "model.layers.1.mlp.experts.16.down_proj.weight", "model.layers.1.mlp.experts.17.gate_and_up_proj.weight", "model.layers.1.mlp.experts.17.down_proj.weight", "model.layers.1.mlp.experts.18.gate_and_up_proj.weight", "model.layers.1.mlp.experts.18.down_proj.weight", "model.layers.1.mlp.experts.19.gate_and_up_proj.weight", "model.layers.1.mlp.experts.19.down_proj.weight", "model.layers.1.mlp.experts.20.gate_and_up_proj.weight", "model.layers.1.mlp.experts.20.down_proj.weight", "model.layers.1.mlp.experts.21.gate_and_up_proj.weight", "model.layers.1.mlp.experts.21.down_proj.weight", "model.layers.1.mlp.experts.22.gate_and_up_proj.weight", "model.layers.1.mlp.experts.22.down_proj.weight", "model.layers.1.mlp.experts.23.gate_and_up_proj.weight", "model.layers.1.mlp.experts.23.down_proj.weight", "model.layers.1.mlp.experts.24.gate_and_up_proj.weight", "model.layers.1.mlp.experts.24.down_proj.weight", "model.layers.1.mlp.experts.25.gate_and_up_proj.weight", "model.layers.1.mlp.experts.25.down_proj.weight", "model.layers.1.mlp.experts.26.gate_and_up_proj.weight", "model.layers.1.mlp.experts.26.down_proj.weight", "model.layers.1.mlp.experts.27.gate_and_up_proj.weight", "model.layers.1.mlp.experts.27.down_proj.weight", "model.layers.1.mlp.experts.28.gate_and_up_proj.weight", "model.layers.1.mlp.experts.28.down_proj.weight", "model.layers.1.mlp.experts.29.gate_and_up_proj.weight", "model.layers.1.mlp.experts.29.down_proj.weight", "model.layers.1.mlp.experts.30.gate_and_up_proj.weight", "model.layers.1.mlp.experts.30.down_proj.weight", "model.layers.1.mlp.experts.31.gate_and_up_proj.weight", "model.layers.1.mlp.experts.31.down_proj.weight", "model.layers.1.mlp.experts.32.gate_and_up_proj.weight", "model.layers.1.mlp.experts.32.down_proj.weight", "model.layers.1.mlp.experts.33.gate_and_up_proj.weight", "model.layers.1.mlp.experts.33.down_proj.weight", "model.layers.1.mlp.experts.34.gate_and_up_proj.weight", "model.layers.1.mlp.experts.34.down_proj.weight", "model.layers.1.mlp.experts.35.gate_and_up_proj.weight", "model.layers.1.mlp.experts.35.down_proj.weight", "model.layers.1.mlp.experts.36.gate_and_up_proj.weight", "model.layers.1.mlp.experts.36.down_proj.weight", "model.layers.1.mlp.experts.37.gate_and_up_proj.weight", "model.layers.1.mlp.experts.37.down_proj.weight", "model.layers.1.mlp.experts.38.gate_and_up_proj.weight", "model.layers.1.mlp.experts.38.down_proj.weight", "model.layers.1.mlp.experts.39.gate_and_up_proj.weight", "model.layers.1.mlp.experts.39.down_proj.weight", "model.layers.1.mlp.experts.40.gate_and_up_proj.weight", "model.layers.1.mlp.experts.40.down_proj.weight", "model.layers.1.mlp.experts.41.gate_and_up_proj.weight", "model.layers.1.mlp.experts.41.down_proj.weight", "model.layers.1.mlp.experts.42.gate_and_up_proj.weight", "model.layers.1.mlp.experts.42.down_proj.weight", "model.layers.1.mlp.experts.43.gate_and_up_proj.weight", "model.layers.1.mlp.experts.43.down_proj.weight", "model.layers.1.mlp.experts.44.gate_and_up_proj.weight", "model.layers.1.mlp.experts.44.down_proj.weight", "model.layers.1.mlp.experts.45.gate_and_up_proj.weight", "model.layers.1.mlp.experts.45.down_proj.weight", "model.layers.1.mlp.experts.46.gate_and_up_proj.weight", "model.layers.1.mlp.experts.46.down_proj.weight", "model.layers.1.mlp.experts.47.gate_and_up_proj.weight", "model.layers.1.mlp.experts.47.down_proj.weight", "model.layers.1.mlp.experts.48.gate_and_up_proj.weight", "model.layers.1.mlp.experts.48.down_proj.weight", "model.layers.1.mlp.experts.49.gate_and_up_proj.weight", "model.layers.1.mlp.experts.49.down_proj.weight", "model.layers.1.mlp.experts.50.gate_and_up_proj.weight", "model.layers.1.mlp.experts.50.down_proj.weight", "model.layers.1.mlp.experts.51.gate_and_up_proj.weight", "model.layers.1.mlp.experts.51.down_proj.weight", "model.layers.1.mlp.experts.52.gate_and_up_proj.weight", "model.layers.1.mlp.experts.52.down_proj.weight", "model.layers.1.mlp.experts.53.gate_and_up_proj.weight", "model.layers.1.mlp.experts.53.down_proj.weight", "model.layers.1.mlp.experts.54.gate_and_up_proj.weight", "model.layers.1.mlp.experts.54.down_proj.weight", "model.layers.1.mlp.experts.55.gate_and_up_proj.weight", "model.layers.1.mlp.experts.55.down_proj.weight", "model.layers.1.mlp.experts.56.gate_and_up_proj.weight", "model.layers.1.mlp.experts.56.down_proj.weight", "model.layers.1.mlp.experts.57.gate_and_up_proj.weight", "model.layers.1.mlp.experts.57.down_proj.weight", "model.layers.1.mlp.experts.58.gate_and_up_proj.weight", "model.layers.1.mlp.experts.58.down_proj.weight", "model.layers.1.mlp.experts.59.gate_and_up_proj.weight", "model.layers.1.mlp.experts.59.down_proj.weight", "model.layers.1.mlp.experts.60.gate_and_up_proj.weight", "model.layers.1.mlp.experts.60.down_proj.weight", "model.layers.1.mlp.experts.61.gate_and_up_proj.weight", "model.layers.1.mlp.experts.61.down_proj.weight", "model.layers.1.mlp.experts.62.gate_and_up_proj.weight", "model.layers.1.mlp.experts.62.down_proj.weight", "model.layers.1.mlp.experts.63.gate_and_up_proj.weight", "model.layers.1.mlp.experts.63.down_proj.weight", "model.layers.2.self_attn.qkv_proj.weight", "model.layers.2.self_attn.o_proj.weight", "model.layers.2.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.2.mlp.shared_mlp.down_proj.weight", "model.layers.2.mlp.experts.0.gate_and_up_proj.weight", "model.layers.2.mlp.experts.0.down_proj.weight", "model.layers.2.mlp.experts.1.gate_and_up_proj.weight", "model.layers.2.mlp.experts.1.down_proj.weight", "model.layers.2.mlp.experts.2.gate_and_up_proj.weight", "model.layers.2.mlp.experts.2.down_proj.weight", "model.layers.2.mlp.experts.3.gate_and_up_proj.weight", "model.layers.2.mlp.experts.3.down_proj.weight", "model.layers.2.mlp.experts.4.gate_and_up_proj.weight", "model.layers.2.mlp.experts.4.down_proj.weight", "model.layers.2.mlp.experts.5.gate_and_up_proj.weight", "model.layers.2.mlp.experts.5.down_proj.weight", "model.layers.2.mlp.experts.6.gate_and_up_proj.weight", "model.layers.2.mlp.experts.6.down_proj.weight", "model.layers.2.mlp.experts.7.gate_and_up_proj.weight", "model.layers.2.mlp.experts.7.down_proj.weight", "model.layers.2.mlp.experts.8.gate_and_up_proj.weight", "model.layers.2.mlp.experts.8.down_proj.weight", "model.layers.2.mlp.experts.9.gate_and_up_proj.weight", "model.layers.2.mlp.experts.9.down_proj.weight", "model.layers.2.mlp.experts.10.gate_and_up_proj.weight", "model.layers.2.mlp.experts.10.down_proj.weight", "model.layers.2.mlp.experts.11.gate_and_up_proj.weight", "model.layers.2.mlp.experts.11.down_proj.weight", "model.layers.2.mlp.experts.12.gate_and_up_proj.weight", "model.layers.2.mlp.experts.12.down_proj.weight", "model.layers.2.mlp.experts.13.gate_and_up_proj.weight", "model.layers.2.mlp.experts.13.down_proj.weight", "model.layers.2.mlp.experts.14.gate_and_up_proj.weight", "model.layers.2.mlp.experts.14.down_proj.weight", "model.layers.2.mlp.experts.15.gate_and_up_proj.weight", "model.layers.2.mlp.experts.15.down_proj.weight", "model.layers.2.mlp.experts.16.gate_and_up_proj.weight", "model.layers.2.mlp.experts.16.down_proj.weight", "model.layers.2.mlp.experts.17.gate_and_up_proj.weight", "model.layers.2.mlp.experts.17.down_proj.weight", "model.layers.2.mlp.experts.18.gate_and_up_proj.weight", "model.layers.2.mlp.experts.18.down_proj.weight", "model.layers.2.mlp.experts.19.gate_and_up_proj.weight", "model.layers.2.mlp.experts.19.down_proj.weight", "model.layers.2.mlp.experts.20.gate_and_up_proj.weight", "model.layers.2.mlp.experts.20.down_proj.weight", "model.layers.2.mlp.experts.21.gate_and_up_proj.weight", "model.layers.2.mlp.experts.21.down_proj.weight", "model.layers.2.mlp.experts.22.gate_and_up_proj.weight", "model.layers.2.mlp.experts.22.down_proj.weight", "model.layers.2.mlp.experts.23.gate_and_up_proj.weight", "model.layers.2.mlp.experts.23.down_proj.weight", "model.layers.2.mlp.experts.24.gate_and_up_proj.weight", "model.layers.2.mlp.experts.24.down_proj.weight", "model.layers.2.mlp.experts.25.gate_and_up_proj.weight", "model.layers.2.mlp.experts.25.down_proj.weight", "model.layers.2.mlp.experts.26.gate_and_up_proj.weight", "model.layers.2.mlp.experts.26.down_proj.weight", "model.layers.2.mlp.experts.27.gate_and_up_proj.weight", "model.layers.2.mlp.experts.27.down_proj.weight", "model.layers.2.mlp.experts.28.gate_and_up_proj.weight", "model.layers.2.mlp.experts.28.down_proj.weight", "model.layers.2.mlp.experts.29.gate_and_up_proj.weight", "model.layers.2.mlp.experts.29.down_proj.weight", "model.layers.2.mlp.experts.30.gate_and_up_proj.weight", "model.layers.2.mlp.experts.30.down_proj.weight", "model.layers.2.mlp.experts.31.gate_and_up_proj.weight", "model.layers.2.mlp.experts.31.down_proj.weight", "model.layers.2.mlp.experts.32.gate_and_up_proj.weight", "model.layers.2.mlp.experts.32.down_proj.weight", "model.layers.2.mlp.experts.33.gate_and_up_proj.weight", "model.layers.2.mlp.experts.33.down_proj.weight", "model.layers.2.mlp.experts.34.gate_and_up_proj.weight", "model.layers.2.mlp.experts.34.down_proj.weight", "model.layers.2.mlp.experts.35.gate_and_up_proj.weight", "model.layers.2.mlp.experts.35.down_proj.weight", "model.layers.2.mlp.experts.36.gate_and_up_proj.weight", "model.layers.2.mlp.experts.36.down_proj.weight", "model.layers.2.mlp.experts.37.gate_and_up_proj.weight", "model.layers.2.mlp.experts.37.down_proj.weight", "model.layers.2.mlp.experts.38.gate_and_up_proj.weight", "model.layers.2.mlp.experts.38.down_proj.weight", "model.layers.2.mlp.experts.39.gate_and_up_proj.weight", "model.layers.2.mlp.experts.39.down_proj.weight", "model.layers.2.mlp.experts.40.gate_and_up_proj.weight", "model.layers.2.mlp.experts.40.down_proj.weight", "model.layers.2.mlp.experts.41.gate_and_up_proj.weight", "model.layers.2.mlp.experts.41.down_proj.weight", "model.layers.2.mlp.experts.42.gate_and_up_proj.weight", "model.layers.2.mlp.experts.42.down_proj.weight", "model.layers.2.mlp.experts.43.gate_and_up_proj.weight", "model.layers.2.mlp.experts.43.down_proj.weight", "model.layers.2.mlp.experts.44.gate_and_up_proj.weight", "model.layers.2.mlp.experts.44.down_proj.weight", "model.layers.2.mlp.experts.45.gate_and_up_proj.weight", "model.layers.2.mlp.experts.45.down_proj.weight", "model.layers.2.mlp.experts.46.gate_and_up_proj.weight", "model.layers.2.mlp.experts.46.down_proj.weight", "model.layers.2.mlp.experts.47.gate_and_up_proj.weight", "model.layers.2.mlp.experts.47.down_proj.weight", "model.layers.2.mlp.experts.48.gate_and_up_proj.weight", "model.layers.2.mlp.experts.48.down_proj.weight", "model.layers.2.mlp.experts.49.gate_and_up_proj.weight", "model.layers.2.mlp.experts.49.down_proj.weight", "model.layers.2.mlp.experts.50.gate_and_up_proj.weight", "model.layers.2.mlp.experts.50.down_proj.weight", "model.layers.2.mlp.experts.51.gate_and_up_proj.weight", "model.layers.2.mlp.experts.51.down_proj.weight", "model.layers.2.mlp.experts.52.gate_and_up_proj.weight", "model.layers.2.mlp.experts.52.down_proj.weight", "model.layers.2.mlp.experts.53.gate_and_up_proj.weight", "model.layers.2.mlp.experts.53.down_proj.weight", "model.layers.2.mlp.experts.54.gate_and_up_proj.weight", "model.layers.2.mlp.experts.54.down_proj.weight", "model.layers.2.mlp.experts.55.gate_and_up_proj.weight", "model.layers.2.mlp.experts.55.down_proj.weight", "model.layers.2.mlp.experts.56.gate_and_up_proj.weight", "model.layers.2.mlp.experts.56.down_proj.weight", "model.layers.2.mlp.experts.57.gate_and_up_proj.weight", "model.layers.2.mlp.experts.57.down_proj.weight", "model.layers.2.mlp.experts.58.gate_and_up_proj.weight", "model.layers.2.mlp.experts.58.down_proj.weight", "model.layers.2.mlp.experts.59.gate_and_up_proj.weight", "model.layers.2.mlp.experts.59.down_proj.weight", "model.layers.2.mlp.experts.60.gate_and_up_proj.weight", "model.layers.2.mlp.experts.60.down_proj.weight", "model.layers.2.mlp.experts.61.gate_and_up_proj.weight", "model.layers.2.mlp.experts.61.down_proj.weight", "model.layers.2.mlp.experts.62.gate_and_up_proj.weight", "model.layers.2.mlp.experts.62.down_proj.weight", "model.layers.2.mlp.experts.63.gate_and_up_proj.weight", "model.layers.2.mlp.experts.63.down_proj.weight", "model.layers.3.self_attn.qkv_proj.weight", "model.layers.3.self_attn.o_proj.weight", "model.layers.3.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.3.mlp.shared_mlp.down_proj.weight", "model.layers.3.mlp.experts.0.gate_and_up_proj.weight", "model.layers.3.mlp.experts.0.down_proj.weight", "model.layers.3.mlp.experts.1.gate_and_up_proj.weight", "model.layers.3.mlp.experts.1.down_proj.weight", "model.layers.3.mlp.experts.2.gate_and_up_proj.weight", "model.layers.3.mlp.experts.2.down_proj.weight", "model.layers.3.mlp.experts.3.gate_and_up_proj.weight", "model.layers.3.mlp.experts.3.down_proj.weight", "model.layers.3.mlp.experts.4.gate_and_up_proj.weight", "model.layers.3.mlp.experts.4.down_proj.weight", "model.layers.3.mlp.experts.5.gate_and_up_proj.weight", "model.layers.3.mlp.experts.5.down_proj.weight", "model.layers.3.mlp.experts.6.gate_and_up_proj.weight", "model.layers.3.mlp.experts.6.down_proj.weight", "model.layers.3.mlp.experts.7.gate_and_up_proj.weight", "model.layers.3.mlp.experts.7.down_proj.weight", "model.layers.3.mlp.experts.8.gate_and_up_proj.weight", "model.layers.3.mlp.experts.8.down_proj.weight", "model.layers.3.mlp.experts.9.gate_and_up_proj.weight", "model.layers.3.mlp.experts.9.down_proj.weight", "model.layers.3.mlp.experts.10.gate_and_up_proj.weight", "model.layers.3.mlp.experts.10.down_proj.weight", "model.layers.3.mlp.experts.11.gate_and_up_proj.weight", "model.layers.3.mlp.experts.11.down_proj.weight", "model.layers.3.mlp.experts.12.gate_and_up_proj.weight", "model.layers.3.mlp.experts.12.down_proj.weight", "model.layers.3.mlp.experts.13.gate_and_up_proj.weight", "model.layers.3.mlp.experts.13.down_proj.weight", "model.layers.3.mlp.experts.14.gate_and_up_proj.weight", "model.layers.3.mlp.experts.14.down_proj.weight", "model.layers.3.mlp.experts.15.gate_and_up_proj.weight", "model.layers.3.mlp.experts.15.down_proj.weight", "model.layers.3.mlp.experts.16.gate_and_up_proj.weight", "model.layers.3.mlp.experts.16.down_proj.weight", "model.layers.3.mlp.experts.17.gate_and_up_proj.weight", "model.layers.3.mlp.experts.17.down_proj.weight", "model.layers.3.mlp.experts.18.gate_and_up_proj.weight", "model.layers.3.mlp.experts.18.down_proj.weight", "model.layers.3.mlp.experts.19.gate_and_up_proj.weight", "model.layers.3.mlp.experts.19.down_proj.weight", "model.layers.3.mlp.experts.20.gate_and_up_proj.weight", "model.layers.3.mlp.experts.20.down_proj.weight", "model.layers.3.mlp.experts.21.gate_and_up_proj.weight", "model.layers.3.mlp.experts.21.down_proj.weight", "model.layers.3.mlp.experts.22.gate_and_up_proj.weight", "model.layers.3.mlp.experts.22.down_proj.weight", "model.layers.3.mlp.experts.23.gate_and_up_proj.weight", "model.layers.3.mlp.experts.23.down_proj.weight", "model.layers.3.mlp.experts.24.gate_and_up_proj.weight", "model.layers.3.mlp.experts.24.down_proj.weight", "model.layers.3.mlp.experts.25.gate_and_up_proj.weight", "model.layers.3.mlp.experts.25.down_proj.weight", "model.layers.3.mlp.experts.26.gate_and_up_proj.weight", "model.layers.3.mlp.experts.26.down_proj.weight", "model.layers.3.mlp.experts.27.gate_and_up_proj.weight", "model.layers.3.mlp.experts.27.down_proj.weight", "model.layers.3.mlp.experts.28.gate_and_up_proj.weight", "model.layers.3.mlp.experts.28.down_proj.weight", "model.layers.3.mlp.experts.29.gate_and_up_proj.weight", "model.layers.3.mlp.experts.29.down_proj.weight", "model.layers.3.mlp.experts.30.gate_and_up_proj.weight", "model.layers.3.mlp.experts.30.down_proj.weight", "model.layers.3.mlp.experts.31.gate_and_up_proj.weight", "model.layers.3.mlp.experts.31.down_proj.weight", "model.layers.3.mlp.experts.32.gate_and_up_proj.weight", "model.layers.3.mlp.experts.32.down_proj.weight", "model.layers.3.mlp.experts.33.gate_and_up_proj.weight", "model.layers.3.mlp.experts.33.down_proj.weight", "model.layers.3.mlp.experts.34.gate_and_up_proj.weight", "model.layers.3.mlp.experts.34.down_proj.weight", "model.layers.3.mlp.experts.35.gate_and_up_proj.weight", "model.layers.3.mlp.experts.35.down_proj.weight", "model.layers.3.mlp.experts.36.gate_and_up_proj.weight", "model.layers.3.mlp.experts.36.down_proj.weight", "model.layers.3.mlp.experts.37.gate_and_up_proj.weight", "model.layers.3.mlp.experts.37.down_proj.weight", "model.layers.3.mlp.experts.38.gate_and_up_proj.weight", "model.layers.3.mlp.experts.38.down_proj.weight", "model.layers.3.mlp.experts.39.gate_and_up_proj.weight", "model.layers.3.mlp.experts.39.down_proj.weight", "model.layers.3.mlp.experts.40.gate_and_up_proj.weight", "model.layers.3.mlp.experts.40.down_proj.weight", "model.layers.3.mlp.experts.41.gate_and_up_proj.weight", "model.layers.3.mlp.experts.41.down_proj.weight", "model.layers.3.mlp.experts.42.gate_and_up_proj.weight", "model.layers.3.mlp.experts.42.down_proj.weight", "model.layers.3.mlp.experts.43.gate_and_up_proj.weight", "model.layers.3.mlp.experts.43.down_proj.weight", "model.layers.3.mlp.experts.44.gate_and_up_proj.weight", "model.layers.3.mlp.experts.44.down_proj.weight", "model.layers.3.mlp.experts.45.gate_and_up_proj.weight", "model.layers.3.mlp.experts.45.down_proj.weight", "model.layers.3.mlp.experts.46.gate_and_up_proj.weight", "model.layers.3.mlp.experts.46.down_proj.weight", "model.layers.3.mlp.experts.47.gate_and_up_proj.weight", "model.layers.3.mlp.experts.47.down_proj.weight", "model.layers.3.mlp.experts.48.gate_and_up_proj.weight", "model.layers.3.mlp.experts.48.down_proj.weight", "model.layers.3.mlp.experts.49.gate_and_up_proj.weight", "model.layers.3.mlp.experts.49.down_proj.weight", "model.layers.3.mlp.experts.50.gate_and_up_proj.weight", "model.layers.3.mlp.experts.50.down_proj.weight", "model.layers.3.mlp.experts.51.gate_and_up_proj.weight", "model.layers.3.mlp.experts.51.down_proj.weight", "model.layers.3.mlp.experts.52.gate_and_up_proj.weight", "model.layers.3.mlp.experts.52.down_proj.weight", "model.layers.3.mlp.experts.53.gate_and_up_proj.weight", "model.layers.3.mlp.experts.53.down_proj.weight", "model.layers.3.mlp.experts.54.gate_and_up_proj.weight", "model.layers.3.mlp.experts.54.down_proj.weight", "model.layers.3.mlp.experts.55.gate_and_up_proj.weight", "model.layers.3.mlp.experts.55.down_proj.weight", "model.layers.3.mlp.experts.56.gate_and_up_proj.weight", "model.layers.3.mlp.experts.56.down_proj.weight", "model.layers.3.mlp.experts.57.gate_and_up_proj.weight", "model.layers.3.mlp.experts.57.down_proj.weight", "model.layers.3.mlp.experts.58.gate_and_up_proj.weight", "model.layers.3.mlp.experts.58.down_proj.weight", "model.layers.3.mlp.experts.59.gate_and_up_proj.weight", "model.layers.3.mlp.experts.59.down_proj.weight", "model.layers.3.mlp.experts.60.gate_and_up_proj.weight", "model.layers.3.mlp.experts.60.down_proj.weight", "model.layers.3.mlp.experts.61.gate_and_up_proj.weight", "model.layers.3.mlp.experts.61.down_proj.weight", "model.layers.3.mlp.experts.62.gate_and_up_proj.weight", "model.layers.3.mlp.experts.62.down_proj.weight", "model.layers.3.mlp.experts.63.gate_and_up_proj.weight", "model.layers.3.mlp.experts.63.down_proj.weight", "model.layers.4.self_attn.qkv_proj.weight", "model.layers.4.self_attn.o_proj.weight", "model.layers.4.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.4.mlp.shared_mlp.down_proj.weight", "model.layers.4.mlp.experts.0.gate_and_up_proj.weight", "model.layers.4.mlp.experts.0.down_proj.weight", "model.layers.4.mlp.experts.1.gate_and_up_proj.weight", "model.layers.4.mlp.experts.1.down_proj.weight", "model.layers.4.mlp.experts.2.gate_and_up_proj.weight", "model.layers.4.mlp.experts.2.down_proj.weight", "model.layers.4.mlp.experts.3.gate_and_up_proj.weight", "model.layers.4.mlp.experts.3.down_proj.weight", "model.layers.4.mlp.experts.4.gate_and_up_proj.weight", "model.layers.4.mlp.experts.4.down_proj.weight", "model.layers.4.mlp.experts.5.gate_and_up_proj.weight", "model.layers.4.mlp.experts.5.down_proj.weight", "model.layers.4.mlp.experts.6.gate_and_up_proj.weight", "model.layers.4.mlp.experts.6.down_proj.weight", "model.layers.4.mlp.experts.7.gate_and_up_proj.weight", "model.layers.4.mlp.experts.7.down_proj.weight", "model.layers.4.mlp.experts.8.gate_and_up_proj.weight", "model.layers.4.mlp.experts.8.down_proj.weight", "model.layers.4.mlp.experts.9.gate_and_up_proj.weight", "model.layers.4.mlp.experts.9.down_proj.weight", "model.layers.4.mlp.experts.10.gate_and_up_proj.weight", "model.layers.4.mlp.experts.10.down_proj.weight", "model.layers.4.mlp.experts.11.gate_and_up_proj.weight", "model.layers.4.mlp.experts.11.down_proj.weight", "model.layers.4.mlp.experts.12.gate_and_up_proj.weight", "model.layers.4.mlp.experts.12.down_proj.weight", "model.layers.4.mlp.experts.13.gate_and_up_proj.weight", "model.layers.4.mlp.experts.13.down_proj.weight", "model.layers.4.mlp.experts.14.gate_and_up_proj.weight", "model.layers.4.mlp.experts.14.down_proj.weight", "model.layers.4.mlp.experts.15.gate_and_up_proj.weight", "model.layers.4.mlp.experts.15.down_proj.weight", "model.layers.4.mlp.experts.16.gate_and_up_proj.weight", "model.layers.4.mlp.experts.16.down_proj.weight", "model.layers.4.mlp.experts.17.gate_and_up_proj.weight", "model.layers.4.mlp.experts.17.down_proj.weight", "model.layers.4.mlp.experts.18.gate_and_up_proj.weight", "model.layers.4.mlp.experts.18.down_proj.weight", "model.layers.4.mlp.experts.19.gate_and_up_proj.weight", "model.layers.4.mlp.experts.19.down_proj.weight", "model.layers.4.mlp.experts.20.gate_and_up_proj.weight", "model.layers.4.mlp.experts.20.down_proj.weight", "model.layers.4.mlp.experts.21.gate_and_up_proj.weight", "model.layers.4.mlp.experts.21.down_proj.weight", "model.layers.4.mlp.experts.22.gate_and_up_proj.weight", "model.layers.4.mlp.experts.22.down_proj.weight", "model.layers.4.mlp.experts.23.gate_and_up_proj.weight", "model.layers.4.mlp.experts.23.down_proj.weight", "model.layers.4.mlp.experts.24.gate_and_up_proj.weight", "model.layers.4.mlp.experts.24.down_proj.weight", "model.layers.4.mlp.experts.25.gate_and_up_proj.weight", "model.layers.4.mlp.experts.25.down_proj.weight", "model.layers.4.mlp.experts.26.gate_and_up_proj.weight", "model.layers.4.mlp.experts.26.down_proj.weight", "model.layers.4.mlp.experts.27.gate_and_up_proj.weight", "model.layers.4.mlp.experts.27.down_proj.weight", "model.layers.4.mlp.experts.28.gate_and_up_proj.weight", "model.layers.4.mlp.experts.28.down_proj.weight", "model.layers.4.mlp.experts.29.gate_and_up_proj.weight", "model.layers.4.mlp.experts.29.down_proj.weight", "model.layers.4.mlp.experts.30.gate_and_up_proj.weight", "model.layers.4.mlp.experts.30.down_proj.weight", "model.layers.4.mlp.experts.31.gate_and_up_proj.weight", "model.layers.4.mlp.experts.31.down_proj.weight", "model.layers.4.mlp.experts.32.gate_and_up_proj.weight", "model.layers.4.mlp.experts.32.down_proj.weight", "model.layers.4.mlp.experts.33.gate_and_up_proj.weight", "model.layers.4.mlp.experts.33.down_proj.weight", "model.layers.4.mlp.experts.34.gate_and_up_proj.weight", "model.layers.4.mlp.experts.34.down_proj.weight", "model.layers.4.mlp.experts.35.gate_and_up_proj.weight", "model.layers.4.mlp.experts.35.down_proj.weight", "model.layers.4.mlp.experts.36.gate_and_up_proj.weight", "model.layers.4.mlp.experts.36.down_proj.weight", "model.layers.4.mlp.experts.37.gate_and_up_proj.weight", "model.layers.4.mlp.experts.37.down_proj.weight", "model.layers.4.mlp.experts.38.gate_and_up_proj.weight", "model.layers.4.mlp.experts.38.down_proj.weight", "model.layers.4.mlp.experts.39.gate_and_up_proj.weight", "model.layers.4.mlp.experts.39.down_proj.weight", "model.layers.4.mlp.experts.40.gate_and_up_proj.weight", "model.layers.4.mlp.experts.40.down_proj.weight", "model.layers.4.mlp.experts.41.gate_and_up_proj.weight", "model.layers.4.mlp.experts.41.down_proj.weight", "model.layers.4.mlp.experts.42.gate_and_up_proj.weight", "model.layers.4.mlp.experts.42.down_proj.weight", "model.layers.4.mlp.experts.43.gate_and_up_proj.weight", "model.layers.4.mlp.experts.43.down_proj.weight", "model.layers.4.mlp.experts.44.gate_and_up_proj.weight", "model.layers.4.mlp.experts.44.down_proj.weight", "model.layers.4.mlp.experts.45.gate_and_up_proj.weight", "model.layers.4.mlp.experts.45.down_proj.weight", "model.layers.4.mlp.experts.46.gate_and_up_proj.weight", "model.layers.4.mlp.experts.46.down_proj.weight", "model.layers.4.mlp.experts.47.gate_and_up_proj.weight", "model.layers.4.mlp.experts.47.down_proj.weight", "model.layers.4.mlp.experts.48.gate_and_up_proj.weight", "model.layers.4.mlp.experts.48.down_proj.weight", "model.layers.4.mlp.experts.49.gate_and_up_proj.weight", "model.layers.4.mlp.experts.49.down_proj.weight", "model.layers.4.mlp.experts.50.gate_and_up_proj.weight", "model.layers.4.mlp.experts.50.down_proj.weight", "model.layers.4.mlp.experts.51.gate_and_up_proj.weight", "model.layers.4.mlp.experts.51.down_proj.weight", "model.layers.4.mlp.experts.52.gate_and_up_proj.weight", "model.layers.4.mlp.experts.52.down_proj.weight", "model.layers.4.mlp.experts.53.gate_and_up_proj.weight", "model.layers.4.mlp.experts.53.down_proj.weight", "model.layers.4.mlp.experts.54.gate_and_up_proj.weight", "model.layers.4.mlp.experts.54.down_proj.weight", "model.layers.4.mlp.experts.55.gate_and_up_proj.weight", "model.layers.4.mlp.experts.55.down_proj.weight", "model.layers.4.mlp.experts.56.gate_and_up_proj.weight", "model.layers.4.mlp.experts.56.down_proj.weight", "model.layers.4.mlp.experts.57.gate_and_up_proj.weight", "model.layers.4.mlp.experts.57.down_proj.weight", "model.layers.4.mlp.experts.58.gate_and_up_proj.weight", "model.layers.4.mlp.experts.58.down_proj.weight", "model.layers.4.mlp.experts.59.gate_and_up_proj.weight", "model.layers.4.mlp.experts.59.down_proj.weight", "model.layers.4.mlp.experts.60.gate_and_up_proj.weight", "model.layers.4.mlp.experts.60.down_proj.weight", "model.layers.4.mlp.experts.61.gate_and_up_proj.weight", "model.layers.4.mlp.experts.61.down_proj.weight", "model.layers.4.mlp.experts.62.gate_and_up_proj.weight", "model.layers.4.mlp.experts.62.down_proj.weight", "model.layers.4.mlp.experts.63.gate_and_up_proj.weight", "model.layers.4.mlp.experts.63.down_proj.weight", "model.layers.5.self_attn.qkv_proj.weight", "model.layers.5.self_attn.o_proj.weight", "model.layers.5.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.5.mlp.shared_mlp.down_proj.weight", "model.layers.5.mlp.experts.0.gate_and_up_proj.weight", "model.layers.5.mlp.experts.0.down_proj.weight", "model.layers.5.mlp.experts.1.gate_and_up_proj.weight", "model.layers.5.mlp.experts.1.down_proj.weight", "model.layers.5.mlp.experts.2.gate_and_up_proj.weight", "model.layers.5.mlp.experts.2.down_proj.weight", "model.layers.5.mlp.experts.3.gate_and_up_proj.weight", "model.layers.5.mlp.experts.3.down_proj.weight", "model.layers.5.mlp.experts.4.gate_and_up_proj.weight", "model.layers.5.mlp.experts.4.down_proj.weight", "model.layers.5.mlp.experts.5.gate_and_up_proj.weight", "model.layers.5.mlp.experts.5.down_proj.weight", "model.layers.5.mlp.experts.6.gate_and_up_proj.weight", "model.layers.5.mlp.experts.6.down_proj.weight", "model.layers.5.mlp.experts.7.gate_and_up_proj.weight", "model.layers.5.mlp.experts.7.down_proj.weight", "model.layers.5.mlp.experts.8.gate_and_up_proj.weight", "model.layers.5.mlp.experts.8.down_proj.weight", "model.layers.5.mlp.experts.9.gate_and_up_proj.weight", "model.layers.5.mlp.experts.9.down_proj.weight", "model.layers.5.mlp.experts.10.gate_and_up_proj.weight", "model.layers.5.mlp.experts.10.down_proj.weight", "model.layers.5.mlp.experts.11.gate_and_up_proj.weight", "model.layers.5.mlp.experts.11.down_proj.weight", "model.layers.5.mlp.experts.12.gate_and_up_proj.weight", "model.layers.5.mlp.experts.12.down_proj.weight", "model.layers.5.mlp.experts.13.gate_and_up_proj.weight", "model.layers.5.mlp.experts.13.down_proj.weight", "model.layers.5.mlp.experts.14.gate_and_up_proj.weight", "model.layers.5.mlp.experts.14.down_proj.weight", "model.layers.5.mlp.experts.15.gate_and_up_proj.weight", "model.layers.5.mlp.experts.15.down_proj.weight", "model.layers.5.mlp.experts.16.gate_and_up_proj.weight", "model.layers.5.mlp.experts.16.down_proj.weight", "model.layers.5.mlp.experts.17.gate_and_up_proj.weight", "model.layers.5.mlp.experts.17.down_proj.weight", "model.layers.5.mlp.experts.18.gate_and_up_proj.weight", "model.layers.5.mlp.experts.18.down_proj.weight", "model.layers.5.mlp.experts.19.gate_and_up_proj.weight", "model.layers.5.mlp.experts.19.down_proj.weight", "model.layers.5.mlp.experts.20.gate_and_up_proj.weight", "model.layers.5.mlp.experts.20.down_proj.weight", "model.layers.5.mlp.experts.21.gate_and_up_proj.weight", "model.layers.5.mlp.experts.21.down_proj.weight", "model.layers.5.mlp.experts.22.gate_and_up_proj.weight", "model.layers.5.mlp.experts.22.down_proj.weight", "model.layers.5.mlp.experts.23.gate_and_up_proj.weight", "model.layers.5.mlp.experts.23.down_proj.weight", "model.layers.5.mlp.experts.24.gate_and_up_proj.weight", "model.layers.5.mlp.experts.24.down_proj.weight", "model.layers.5.mlp.experts.25.gate_and_up_proj.weight", "model.layers.5.mlp.experts.25.down_proj.weight", "model.layers.5.mlp.experts.26.gate_and_up_proj.weight", "model.layers.5.mlp.experts.26.down_proj.weight", "model.layers.5.mlp.experts.27.gate_and_up_proj.weight", "model.layers.5.mlp.experts.27.down_proj.weight", "model.layers.5.mlp.experts.28.gate_and_up_proj.weight", "model.layers.5.mlp.experts.28.down_proj.weight", "model.layers.5.mlp.experts.29.gate_and_up_proj.weight", "model.layers.5.mlp.experts.29.down_proj.weight", "model.layers.5.mlp.experts.30.gate_and_up_proj.weight", "model.layers.5.mlp.experts.30.down_proj.weight", "model.layers.5.mlp.experts.31.gate_and_up_proj.weight", "model.layers.5.mlp.experts.31.down_proj.weight", "model.layers.5.mlp.experts.32.gate_and_up_proj.weight", "model.layers.5.mlp.experts.32.down_proj.weight", "model.layers.5.mlp.experts.33.gate_and_up_proj.weight", "model.layers.5.mlp.experts.33.down_proj.weight", "model.layers.5.mlp.experts.34.gate_and_up_proj.weight", "model.layers.5.mlp.experts.34.down_proj.weight", "model.layers.5.mlp.experts.35.gate_and_up_proj.weight", "model.layers.5.mlp.experts.35.down_proj.weight", "model.layers.5.mlp.experts.36.gate_and_up_proj.weight", "model.layers.5.mlp.experts.36.down_proj.weight", "model.layers.5.mlp.experts.37.gate_and_up_proj.weight", "model.layers.5.mlp.experts.37.down_proj.weight", "model.layers.5.mlp.experts.38.gate_and_up_proj.weight", "model.layers.5.mlp.experts.38.down_proj.weight", "model.layers.5.mlp.experts.39.gate_and_up_proj.weight", "model.layers.5.mlp.experts.39.down_proj.weight", "model.layers.5.mlp.experts.40.gate_and_up_proj.weight", "model.layers.5.mlp.experts.40.down_proj.weight", "model.layers.5.mlp.experts.41.gate_and_up_proj.weight", "model.layers.5.mlp.experts.41.down_proj.weight", "model.layers.5.mlp.experts.42.gate_and_up_proj.weight", "model.layers.5.mlp.experts.42.down_proj.weight", "model.layers.5.mlp.experts.43.gate_and_up_proj.weight", "model.layers.5.mlp.experts.43.down_proj.weight", "model.layers.5.mlp.experts.44.gate_and_up_proj.weight", "model.layers.5.mlp.experts.44.down_proj.weight", "model.layers.5.mlp.experts.45.gate_and_up_proj.weight", "model.layers.5.mlp.experts.45.down_proj.weight", "model.layers.5.mlp.experts.46.gate_and_up_proj.weight", "model.layers.5.mlp.experts.46.down_proj.weight", "model.layers.5.mlp.experts.47.gate_and_up_proj.weight", "model.layers.5.mlp.experts.47.down_proj.weight", "model.layers.5.mlp.experts.48.gate_and_up_proj.weight", "model.layers.5.mlp.experts.48.down_proj.weight", "model.layers.5.mlp.experts.49.gate_and_up_proj.weight", "model.layers.5.mlp.experts.49.down_proj.weight", "model.layers.5.mlp.experts.50.gate_and_up_proj.weight", "model.layers.5.mlp.experts.50.down_proj.weight", "model.layers.5.mlp.experts.51.gate_and_up_proj.weight", "model.layers.5.mlp.experts.51.down_proj.weight", "model.layers.5.mlp.experts.52.gate_and_up_proj.weight", "model.layers.5.mlp.experts.52.down_proj.weight", "model.layers.5.mlp.experts.53.gate_and_up_proj.weight", "model.layers.5.mlp.experts.53.down_proj.weight", "model.layers.5.mlp.experts.54.gate_and_up_proj.weight", "model.layers.5.mlp.experts.54.down_proj.weight", "model.layers.5.mlp.experts.55.gate_and_up_proj.weight", "model.layers.5.mlp.experts.55.down_proj.weight", "model.layers.5.mlp.experts.56.gate_and_up_proj.weight", "model.layers.5.mlp.experts.56.down_proj.weight", "model.layers.5.mlp.experts.57.gate_and_up_proj.weight", "model.layers.5.mlp.experts.57.down_proj.weight", "model.layers.5.mlp.experts.58.gate_and_up_proj.weight", "model.layers.5.mlp.experts.58.down_proj.weight", "model.layers.5.mlp.experts.59.gate_and_up_proj.weight", "model.layers.5.mlp.experts.59.down_proj.weight", "model.layers.5.mlp.experts.60.gate_and_up_proj.weight", "model.layers.5.mlp.experts.60.down_proj.weight", "model.layers.5.mlp.experts.61.gate_and_up_proj.weight", "model.layers.5.mlp.experts.61.down_proj.weight", "model.layers.5.mlp.experts.62.gate_and_up_proj.weight", "model.layers.5.mlp.experts.62.down_proj.weight", "model.layers.5.mlp.experts.63.gate_and_up_proj.weight", "model.layers.5.mlp.experts.63.down_proj.weight", "model.layers.6.self_attn.qkv_proj.weight", "model.layers.6.self_attn.o_proj.weight", "model.layers.6.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.6.mlp.shared_mlp.down_proj.weight", "model.layers.6.mlp.experts.0.gate_and_up_proj.weight", "model.layers.6.mlp.experts.0.down_proj.weight", "model.layers.6.mlp.experts.1.gate_and_up_proj.weight", "model.layers.6.mlp.experts.1.down_proj.weight", "model.layers.6.mlp.experts.2.gate_and_up_proj.weight", "model.layers.6.mlp.experts.2.down_proj.weight", "model.layers.6.mlp.experts.3.gate_and_up_proj.weight", "model.layers.6.mlp.experts.3.down_proj.weight", "model.layers.6.mlp.experts.4.gate_and_up_proj.weight", "model.layers.6.mlp.experts.4.down_proj.weight", "model.layers.6.mlp.experts.5.gate_and_up_proj.weight", "model.layers.6.mlp.experts.5.down_proj.weight", "model.layers.6.mlp.experts.6.gate_and_up_proj.weight", "model.layers.6.mlp.experts.6.down_proj.weight", "model.layers.6.mlp.experts.7.gate_and_up_proj.weight", "model.layers.6.mlp.experts.7.down_proj.weight", "model.layers.6.mlp.experts.8.gate_and_up_proj.weight", "model.layers.6.mlp.experts.8.down_proj.weight", "model.layers.6.mlp.experts.9.gate_and_up_proj.weight", "model.layers.6.mlp.experts.9.down_proj.weight", "model.layers.6.mlp.experts.10.gate_and_up_proj.weight", "model.layers.6.mlp.experts.10.down_proj.weight", "model.layers.6.mlp.experts.11.gate_and_up_proj.weight", "model.layers.6.mlp.experts.11.down_proj.weight", "model.layers.6.mlp.experts.12.gate_and_up_proj.weight", "model.layers.6.mlp.experts.12.down_proj.weight", "model.layers.6.mlp.experts.13.gate_and_up_proj.weight", "model.layers.6.mlp.experts.13.down_proj.weight", "model.layers.6.mlp.experts.14.gate_and_up_proj.weight", "model.layers.6.mlp.experts.14.down_proj.weight", "model.layers.6.mlp.experts.15.gate_and_up_proj.weight", "model.layers.6.mlp.experts.15.down_proj.weight", "model.layers.6.mlp.experts.16.gate_and_up_proj.weight", "model.layers.6.mlp.experts.16.down_proj.weight", "model.layers.6.mlp.experts.17.gate_and_up_proj.weight", "model.layers.6.mlp.experts.17.down_proj.weight", "model.layers.6.mlp.experts.18.gate_and_up_proj.weight", "model.layers.6.mlp.experts.18.down_proj.weight", "model.layers.6.mlp.experts.19.gate_and_up_proj.weight", "model.layers.6.mlp.experts.19.down_proj.weight", "model.layers.6.mlp.experts.20.gate_and_up_proj.weight", "model.layers.6.mlp.experts.20.down_proj.weight", "model.layers.6.mlp.experts.21.gate_and_up_proj.weight", "model.layers.6.mlp.experts.21.down_proj.weight", "model.layers.6.mlp.experts.22.gate_and_up_proj.weight", "model.layers.6.mlp.experts.22.down_proj.weight", "model.layers.6.mlp.experts.23.gate_and_up_proj.weight", "model.layers.6.mlp.experts.23.down_proj.weight", "model.layers.6.mlp.experts.24.gate_and_up_proj.weight", "model.layers.6.mlp.experts.24.down_proj.weight", "model.layers.6.mlp.experts.25.gate_and_up_proj.weight", "model.layers.6.mlp.experts.25.down_proj.weight", "model.layers.6.mlp.experts.26.gate_and_up_proj.weight", "model.layers.6.mlp.experts.26.down_proj.weight", "model.layers.6.mlp.experts.27.gate_and_up_proj.weight", "model.layers.6.mlp.experts.27.down_proj.weight", "model.layers.6.mlp.experts.28.gate_and_up_proj.weight", "model.layers.6.mlp.experts.28.down_proj.weight", "model.layers.6.mlp.experts.29.gate_and_up_proj.weight", "model.layers.6.mlp.experts.29.down_proj.weight", "model.layers.6.mlp.experts.30.gate_and_up_proj.weight", "model.layers.6.mlp.experts.30.down_proj.weight", "model.layers.6.mlp.experts.31.gate_and_up_proj.weight", "model.layers.6.mlp.experts.31.down_proj.weight", "model.layers.6.mlp.experts.32.gate_and_up_proj.weight", "model.layers.6.mlp.experts.32.down_proj.weight", "model.layers.6.mlp.experts.33.gate_and_up_proj.weight", "model.layers.6.mlp.experts.33.down_proj.weight", "model.layers.6.mlp.experts.34.gate_and_up_proj.weight", "model.layers.6.mlp.experts.34.down_proj.weight", "model.layers.6.mlp.experts.35.gate_and_up_proj.weight", "model.layers.6.mlp.experts.35.down_proj.weight", "model.layers.6.mlp.experts.36.gate_and_up_proj.weight", "model.layers.6.mlp.experts.36.down_proj.weight", "model.layers.6.mlp.experts.37.gate_and_up_proj.weight", "model.layers.6.mlp.experts.37.down_proj.weight", "model.layers.6.mlp.experts.38.gate_and_up_proj.weight", "model.layers.6.mlp.experts.38.down_proj.weight", "model.layers.6.mlp.experts.39.gate_and_up_proj.weight", "model.layers.6.mlp.experts.39.down_proj.weight", "model.layers.6.mlp.experts.40.gate_and_up_proj.weight", "model.layers.6.mlp.experts.40.down_proj.weight", "model.layers.6.mlp.experts.41.gate_and_up_proj.weight", "model.layers.6.mlp.experts.41.down_proj.weight", "model.layers.6.mlp.experts.42.gate_and_up_proj.weight", "model.layers.6.mlp.experts.42.down_proj.weight", "model.layers.6.mlp.experts.43.gate_and_up_proj.weight", "model.layers.6.mlp.experts.43.down_proj.weight", "model.layers.6.mlp.experts.44.gate_and_up_proj.weight", "model.layers.6.mlp.experts.44.down_proj.weight", "model.layers.6.mlp.experts.45.gate_and_up_proj.weight", "model.layers.6.mlp.experts.45.down_proj.weight", "model.layers.6.mlp.experts.46.gate_and_up_proj.weight", "model.layers.6.mlp.experts.46.down_proj.weight", "model.layers.6.mlp.experts.47.gate_and_up_proj.weight", "model.layers.6.mlp.experts.47.down_proj.weight", "model.layers.6.mlp.experts.48.gate_and_up_proj.weight", "model.layers.6.mlp.experts.48.down_proj.weight", "model.layers.6.mlp.experts.49.gate_and_up_proj.weight", "model.layers.6.mlp.experts.49.down_proj.weight", "model.layers.6.mlp.experts.50.gate_and_up_proj.weight", "model.layers.6.mlp.experts.50.down_proj.weight", "model.layers.6.mlp.experts.51.gate_and_up_proj.weight", "model.layers.6.mlp.experts.51.down_proj.weight", "model.layers.6.mlp.experts.52.gate_and_up_proj.weight", "model.layers.6.mlp.experts.52.down_proj.weight", "model.layers.6.mlp.experts.53.gate_and_up_proj.weight", "model.layers.6.mlp.experts.53.down_proj.weight", "model.layers.6.mlp.experts.54.gate_and_up_proj.weight", "model.layers.6.mlp.experts.54.down_proj.weight", "model.layers.6.mlp.experts.55.gate_and_up_proj.weight", "model.layers.6.mlp.experts.55.down_proj.weight", "model.layers.6.mlp.experts.56.gate_and_up_proj.weight", "model.layers.6.mlp.experts.56.down_proj.weight", "model.layers.6.mlp.experts.57.gate_and_up_proj.weight", "model.layers.6.mlp.experts.57.down_proj.weight", "model.layers.6.mlp.experts.58.gate_and_up_proj.weight", "model.layers.6.mlp.experts.58.down_proj.weight", "model.layers.6.mlp.experts.59.gate_and_up_proj.weight", "model.layers.6.mlp.experts.59.down_proj.weight", "model.layers.6.mlp.experts.60.gate_and_up_proj.weight", "model.layers.6.mlp.experts.60.down_proj.weight", "model.layers.6.mlp.experts.61.gate_and_up_proj.weight", "model.layers.6.mlp.experts.61.down_proj.weight", "model.layers.6.mlp.experts.62.gate_and_up_proj.weight", "model.layers.6.mlp.experts.62.down_proj.weight", "model.layers.6.mlp.experts.63.gate_and_up_proj.weight", "model.layers.6.mlp.experts.63.down_proj.weight", "model.layers.7.self_attn.qkv_proj.weight", "model.layers.7.self_attn.o_proj.weight", "model.layers.7.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.7.mlp.shared_mlp.down_proj.weight", "model.layers.7.mlp.experts.0.gate_and_up_proj.weight", "model.layers.7.mlp.experts.0.down_proj.weight", "model.layers.7.mlp.experts.1.gate_and_up_proj.weight", "model.layers.7.mlp.experts.1.down_proj.weight", "model.layers.7.mlp.experts.2.gate_and_up_proj.weight", "model.layers.7.mlp.experts.2.down_proj.weight", "model.layers.7.mlp.experts.3.gate_and_up_proj.weight", "model.layers.7.mlp.experts.3.down_proj.weight", "model.layers.7.mlp.experts.4.gate_and_up_proj.weight", "model.layers.7.mlp.experts.4.down_proj.weight", "model.layers.7.mlp.experts.5.gate_and_up_proj.weight", "model.layers.7.mlp.experts.5.down_proj.weight", "model.layers.7.mlp.experts.6.gate_and_up_proj.weight", "model.layers.7.mlp.experts.6.down_proj.weight", "model.layers.7.mlp.experts.7.gate_and_up_proj.weight", "model.layers.7.mlp.experts.7.down_proj.weight", "model.layers.7.mlp.experts.8.gate_and_up_proj.weight", "model.layers.7.mlp.experts.8.down_proj.weight", "model.layers.7.mlp.experts.9.gate_and_up_proj.weight", "model.layers.7.mlp.experts.9.down_proj.weight", "model.layers.7.mlp.experts.10.gate_and_up_proj.weight", "model.layers.7.mlp.experts.10.down_proj.weight", "model.layers.7.mlp.experts.11.gate_and_up_proj.weight", "model.layers.7.mlp.experts.11.down_proj.weight", "model.layers.7.mlp.experts.12.gate_and_up_proj.weight", "model.layers.7.mlp.experts.12.down_proj.weight", "model.layers.7.mlp.experts.13.gate_and_up_proj.weight", "model.layers.7.mlp.experts.13.down_proj.weight", "model.layers.7.mlp.experts.14.gate_and_up_proj.weight", "model.layers.7.mlp.experts.14.down_proj.weight", "model.layers.7.mlp.experts.15.gate_and_up_proj.weight", "model.layers.7.mlp.experts.15.down_proj.weight", "model.layers.7.mlp.experts.16.gate_and_up_proj.weight", "model.layers.7.mlp.experts.16.down_proj.weight", "model.layers.7.mlp.experts.17.gate_and_up_proj.weight", "model.layers.7.mlp.experts.17.down_proj.weight", "model.layers.7.mlp.experts.18.gate_and_up_proj.weight", "model.layers.7.mlp.experts.18.down_proj.weight", "model.layers.7.mlp.experts.19.gate_and_up_proj.weight", "model.layers.7.mlp.experts.19.down_proj.weight", "model.layers.7.mlp.experts.20.gate_and_up_proj.weight", "model.layers.7.mlp.experts.20.down_proj.weight", "model.layers.7.mlp.experts.21.gate_and_up_proj.weight", "model.layers.7.mlp.experts.21.down_proj.weight", "model.layers.7.mlp.experts.22.gate_and_up_proj.weight", "model.layers.7.mlp.experts.22.down_proj.weight", "model.layers.7.mlp.experts.23.gate_and_up_proj.weight", "model.layers.7.mlp.experts.23.down_proj.weight", "model.layers.7.mlp.experts.24.gate_and_up_proj.weight", "model.layers.7.mlp.experts.24.down_proj.weight", "model.layers.7.mlp.experts.25.gate_and_up_proj.weight", "model.layers.7.mlp.experts.25.down_proj.weight", "model.layers.7.mlp.experts.26.gate_and_up_proj.weight", "model.layers.7.mlp.experts.26.down_proj.weight", "model.layers.7.mlp.experts.27.gate_and_up_proj.weight", "model.layers.7.mlp.experts.27.down_proj.weight", "model.layers.7.mlp.experts.28.gate_and_up_proj.weight", "model.layers.7.mlp.experts.28.down_proj.weight", "model.layers.7.mlp.experts.29.gate_and_up_proj.weight", "model.layers.7.mlp.experts.29.down_proj.weight", "model.layers.7.mlp.experts.30.gate_and_up_proj.weight", "model.layers.7.mlp.experts.30.down_proj.weight", "model.layers.7.mlp.experts.31.gate_and_up_proj.weight", "model.layers.7.mlp.experts.31.down_proj.weight", "model.layers.7.mlp.experts.32.gate_and_up_proj.weight", "model.layers.7.mlp.experts.32.down_proj.weight", "model.layers.7.mlp.experts.33.gate_and_up_proj.weight", "model.layers.7.mlp.experts.33.down_proj.weight", "model.layers.7.mlp.experts.34.gate_and_up_proj.weight", "model.layers.7.mlp.experts.34.down_proj.weight", "model.layers.7.mlp.experts.35.gate_and_up_proj.weight", "model.layers.7.mlp.experts.35.down_proj.weight", "model.layers.7.mlp.experts.36.gate_and_up_proj.weight", "model.layers.7.mlp.experts.36.down_proj.weight", "model.layers.7.mlp.experts.37.gate_and_up_proj.weight", "model.layers.7.mlp.experts.37.down_proj.weight", "model.layers.7.mlp.experts.38.gate_and_up_proj.weight", "model.layers.7.mlp.experts.38.down_proj.weight", "model.layers.7.mlp.experts.39.gate_and_up_proj.weight", "model.layers.7.mlp.experts.39.down_proj.weight", "model.layers.7.mlp.experts.40.gate_and_up_proj.weight", "model.layers.7.mlp.experts.40.down_proj.weight", "model.layers.7.mlp.experts.41.gate_and_up_proj.weight", "model.layers.7.mlp.experts.41.down_proj.weight", "model.layers.7.mlp.experts.42.gate_and_up_proj.weight", "model.layers.7.mlp.experts.42.down_proj.weight", "model.layers.7.mlp.experts.43.gate_and_up_proj.weight", "model.layers.7.mlp.experts.43.down_proj.weight", "model.layers.7.mlp.experts.44.gate_and_up_proj.weight", "model.layers.7.mlp.experts.44.down_proj.weight", "model.layers.7.mlp.experts.45.gate_and_up_proj.weight", "model.layers.7.mlp.experts.45.down_proj.weight", "model.layers.7.mlp.experts.46.gate_and_up_proj.weight", "model.layers.7.mlp.experts.46.down_proj.weight", "model.layers.7.mlp.experts.47.gate_and_up_proj.weight", "model.layers.7.mlp.experts.47.down_proj.weight", "model.layers.7.mlp.experts.48.gate_and_up_proj.weight", "model.layers.7.mlp.experts.48.down_proj.weight", "model.layers.7.mlp.experts.49.gate_and_up_proj.weight", "model.layers.7.mlp.experts.49.down_proj.weight", "model.layers.7.mlp.experts.50.gate_and_up_proj.weight", "model.layers.7.mlp.experts.50.down_proj.weight", "model.layers.7.mlp.experts.51.gate_and_up_proj.weight", "model.layers.7.mlp.experts.51.down_proj.weight", "model.layers.7.mlp.experts.52.gate_and_up_proj.weight", "model.layers.7.mlp.experts.52.down_proj.weight", "model.layers.7.mlp.experts.53.gate_and_up_proj.weight", "model.layers.7.mlp.experts.53.down_proj.weight", "model.layers.7.mlp.experts.54.gate_and_up_proj.weight", "model.layers.7.mlp.experts.54.down_proj.weight", "model.layers.7.mlp.experts.55.gate_and_up_proj.weight", "model.layers.7.mlp.experts.55.down_proj.weight", "model.layers.7.mlp.experts.56.gate_and_up_proj.weight", "model.layers.7.mlp.experts.56.down_proj.weight", "model.layers.7.mlp.experts.57.gate_and_up_proj.weight", "model.layers.7.mlp.experts.57.down_proj.weight", "model.layers.7.mlp.experts.58.gate_and_up_proj.weight", "model.layers.7.mlp.experts.58.down_proj.weight", "model.layers.7.mlp.experts.59.gate_and_up_proj.weight", "model.layers.7.mlp.experts.59.down_proj.weight", "model.layers.7.mlp.experts.60.gate_and_up_proj.weight", "model.layers.7.mlp.experts.60.down_proj.weight", "model.layers.7.mlp.experts.61.gate_and_up_proj.weight", "model.layers.7.mlp.experts.61.down_proj.weight", "model.layers.7.mlp.experts.62.gate_and_up_proj.weight", "model.layers.7.mlp.experts.62.down_proj.weight", "model.layers.7.mlp.experts.63.gate_and_up_proj.weight", "model.layers.7.mlp.experts.63.down_proj.weight", "model.layers.8.self_attn.qkv_proj.weight", "model.layers.8.self_attn.o_proj.weight", "model.layers.8.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.8.mlp.shared_mlp.down_proj.weight", "model.layers.8.mlp.experts.0.gate_and_up_proj.weight", "model.layers.8.mlp.experts.0.down_proj.weight", "model.layers.8.mlp.experts.1.gate_and_up_proj.weight", "model.layers.8.mlp.experts.1.down_proj.weight", "model.layers.8.mlp.experts.2.gate_and_up_proj.weight", "model.layers.8.mlp.experts.2.down_proj.weight", "model.layers.8.mlp.experts.3.gate_and_up_proj.weight", "model.layers.8.mlp.experts.3.down_proj.weight", "model.layers.8.mlp.experts.4.gate_and_up_proj.weight", "model.layers.8.mlp.experts.4.down_proj.weight", "model.layers.8.mlp.experts.5.gate_and_up_proj.weight", "model.layers.8.mlp.experts.5.down_proj.weight", "model.layers.8.mlp.experts.6.gate_and_up_proj.weight", "model.layers.8.mlp.experts.6.down_proj.weight", "model.layers.8.mlp.experts.7.gate_and_up_proj.weight", "model.layers.8.mlp.experts.7.down_proj.weight", "model.layers.8.mlp.experts.8.gate_and_up_proj.weight", "model.layers.8.mlp.experts.8.down_proj.weight", "model.layers.8.mlp.experts.9.gate_and_up_proj.weight", "model.layers.8.mlp.experts.9.down_proj.weight", "model.layers.8.mlp.experts.10.gate_and_up_proj.weight", "model.layers.8.mlp.experts.10.down_proj.weight", "model.layers.8.mlp.experts.11.gate_and_up_proj.weight", "model.layers.8.mlp.experts.11.down_proj.weight", "model.layers.8.mlp.experts.12.gate_and_up_proj.weight", "model.layers.8.mlp.experts.12.down_proj.weight", "model.layers.8.mlp.experts.13.gate_and_up_proj.weight", "model.layers.8.mlp.experts.13.down_proj.weight", "model.layers.8.mlp.experts.14.gate_and_up_proj.weight", "model.layers.8.mlp.experts.14.down_proj.weight", "model.layers.8.mlp.experts.15.gate_and_up_proj.weight", "model.layers.8.mlp.experts.15.down_proj.weight", "model.layers.8.mlp.experts.16.gate_and_up_proj.weight", "model.layers.8.mlp.experts.16.down_proj.weight", "model.layers.8.mlp.experts.17.gate_and_up_proj.weight", "model.layers.8.mlp.experts.17.down_proj.weight", "model.layers.8.mlp.experts.18.gate_and_up_proj.weight", "model.layers.8.mlp.experts.18.down_proj.weight", "model.layers.8.mlp.experts.19.gate_and_up_proj.weight", "model.layers.8.mlp.experts.19.down_proj.weight", "model.layers.8.mlp.experts.20.gate_and_up_proj.weight", "model.layers.8.mlp.experts.20.down_proj.weight", "model.layers.8.mlp.experts.21.gate_and_up_proj.weight", "model.layers.8.mlp.experts.21.down_proj.weight", "model.layers.8.mlp.experts.22.gate_and_up_proj.weight", "model.layers.8.mlp.experts.22.down_proj.weight", "model.layers.8.mlp.experts.23.gate_and_up_proj.weight", "model.layers.8.mlp.experts.23.down_proj.weight", "model.layers.8.mlp.experts.24.gate_and_up_proj.weight", "model.layers.8.mlp.experts.24.down_proj.weight", "model.layers.8.mlp.experts.25.gate_and_up_proj.weight", "model.layers.8.mlp.experts.25.down_proj.weight", "model.layers.8.mlp.experts.26.gate_and_up_proj.weight", "model.layers.8.mlp.experts.26.down_proj.weight", "model.layers.8.mlp.experts.27.gate_and_up_proj.weight", "model.layers.8.mlp.experts.27.down_proj.weight", "model.layers.8.mlp.experts.28.gate_and_up_proj.weight", "model.layers.8.mlp.experts.28.down_proj.weight", "model.layers.8.mlp.experts.29.gate_and_up_proj.weight", "model.layers.8.mlp.experts.29.down_proj.weight", "model.layers.8.mlp.experts.30.gate_and_up_proj.weight", "model.layers.8.mlp.experts.30.down_proj.weight", "model.layers.8.mlp.experts.31.gate_and_up_proj.weight", "model.layers.8.mlp.experts.31.down_proj.weight", "model.layers.8.mlp.experts.32.gate_and_up_proj.weight", "model.layers.8.mlp.experts.32.down_proj.weight", "model.layers.8.mlp.experts.33.gate_and_up_proj.weight", "model.layers.8.mlp.experts.33.down_proj.weight", "model.layers.8.mlp.experts.34.gate_and_up_proj.weight", "model.layers.8.mlp.experts.34.down_proj.weight", "model.layers.8.mlp.experts.35.gate_and_up_proj.weight", "model.layers.8.mlp.experts.35.down_proj.weight", "model.layers.8.mlp.experts.36.gate_and_up_proj.weight", "model.layers.8.mlp.experts.36.down_proj.weight", "model.layers.8.mlp.experts.37.gate_and_up_proj.weight", "model.layers.8.mlp.experts.37.down_proj.weight", "model.layers.8.mlp.experts.38.gate_and_up_proj.weight", "model.layers.8.mlp.experts.38.down_proj.weight", "model.layers.8.mlp.experts.39.gate_and_up_proj.weight", "model.layers.8.mlp.experts.39.down_proj.weight", "model.layers.8.mlp.experts.40.gate_and_up_proj.weight", "model.layers.8.mlp.experts.40.down_proj.weight", "model.layers.8.mlp.experts.41.gate_and_up_proj.weight", "model.layers.8.mlp.experts.41.down_proj.weight", "model.layers.8.mlp.experts.42.gate_and_up_proj.weight", "model.layers.8.mlp.experts.42.down_proj.weight", "model.layers.8.mlp.experts.43.gate_and_up_proj.weight", "model.layers.8.mlp.experts.43.down_proj.weight", "model.layers.8.mlp.experts.44.gate_and_up_proj.weight", "model.layers.8.mlp.experts.44.down_proj.weight", "model.layers.8.mlp.experts.45.gate_and_up_proj.weight", "model.layers.8.mlp.experts.45.down_proj.weight", "model.layers.8.mlp.experts.46.gate_and_up_proj.weight", "model.layers.8.mlp.experts.46.down_proj.weight", "model.layers.8.mlp.experts.47.gate_and_up_proj.weight", "model.layers.8.mlp.experts.47.down_proj.weight", "model.layers.8.mlp.experts.48.gate_and_up_proj.weight", "model.layers.8.mlp.experts.48.down_proj.weight", "model.layers.8.mlp.experts.49.gate_and_up_proj.weight", "model.layers.8.mlp.experts.49.down_proj.weight", "model.layers.8.mlp.experts.50.gate_and_up_proj.weight", "model.layers.8.mlp.experts.50.down_proj.weight", "model.layers.8.mlp.experts.51.gate_and_up_proj.weight", "model.layers.8.mlp.experts.51.down_proj.weight", "model.layers.8.mlp.experts.52.gate_and_up_proj.weight", "model.layers.8.mlp.experts.52.down_proj.weight", "model.layers.8.mlp.experts.53.gate_and_up_proj.weight", "model.layers.8.mlp.experts.53.down_proj.weight", "model.layers.8.mlp.experts.54.gate_and_up_proj.weight", "model.layers.8.mlp.experts.54.down_proj.weight", "model.layers.8.mlp.experts.55.gate_and_up_proj.weight", "model.layers.8.mlp.experts.55.down_proj.weight", "model.layers.8.mlp.experts.56.gate_and_up_proj.weight", "model.layers.8.mlp.experts.56.down_proj.weight", "model.layers.8.mlp.experts.57.gate_and_up_proj.weight", "model.layers.8.mlp.experts.57.down_proj.weight", "model.layers.8.mlp.experts.58.gate_and_up_proj.weight", "model.layers.8.mlp.experts.58.down_proj.weight", "model.layers.8.mlp.experts.59.gate_and_up_proj.weight", "model.layers.8.mlp.experts.59.down_proj.weight", "model.layers.8.mlp.experts.60.gate_and_up_proj.weight", "model.layers.8.mlp.experts.60.down_proj.weight", "model.layers.8.mlp.experts.61.gate_and_up_proj.weight", "model.layers.8.mlp.experts.61.down_proj.weight", "model.layers.8.mlp.experts.62.gate_and_up_proj.weight", "model.layers.8.mlp.experts.62.down_proj.weight", "model.layers.8.mlp.experts.63.gate_and_up_proj.weight", "model.layers.8.mlp.experts.63.down_proj.weight", "model.layers.9.self_attn.qkv_proj.weight", "model.layers.9.self_attn.o_proj.weight", "model.layers.9.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.9.mlp.shared_mlp.down_proj.weight", "model.layers.9.mlp.experts.0.gate_and_up_proj.weight", "model.layers.9.mlp.experts.0.down_proj.weight", "model.layers.9.mlp.experts.1.gate_and_up_proj.weight", "model.layers.9.mlp.experts.1.down_proj.weight", "model.layers.9.mlp.experts.2.gate_and_up_proj.weight", "model.layers.9.mlp.experts.2.down_proj.weight", "model.layers.9.mlp.experts.3.gate_and_up_proj.weight", "model.layers.9.mlp.experts.3.down_proj.weight", "model.layers.9.mlp.experts.4.gate_and_up_proj.weight", "model.layers.9.mlp.experts.4.down_proj.weight", "model.layers.9.mlp.experts.5.gate_and_up_proj.weight", "model.layers.9.mlp.experts.5.down_proj.weight", "model.layers.9.mlp.experts.6.gate_and_up_proj.weight", "model.layers.9.mlp.experts.6.down_proj.weight", "model.layers.9.mlp.experts.7.gate_and_up_proj.weight", "model.layers.9.mlp.experts.7.down_proj.weight", "model.layers.9.mlp.experts.8.gate_and_up_proj.weight", "model.layers.9.mlp.experts.8.down_proj.weight", "model.layers.9.mlp.experts.9.gate_and_up_proj.weight", "model.layers.9.mlp.experts.9.down_proj.weight", "model.layers.9.mlp.experts.10.gate_and_up_proj.weight", "model.layers.9.mlp.experts.10.down_proj.weight", "model.layers.9.mlp.experts.11.gate_and_up_proj.weight", "model.layers.9.mlp.experts.11.down_proj.weight", "model.layers.9.mlp.experts.12.gate_and_up_proj.weight", "model.layers.9.mlp.experts.12.down_proj.weight", "model.layers.9.mlp.experts.13.gate_and_up_proj.weight", "model.layers.9.mlp.experts.13.down_proj.weight", "model.layers.9.mlp.experts.14.gate_and_up_proj.weight", "model.layers.9.mlp.experts.14.down_proj.weight", "model.layers.9.mlp.experts.15.gate_and_up_proj.weight", "model.layers.9.mlp.experts.15.down_proj.weight", "model.layers.9.mlp.experts.16.gate_and_up_proj.weight", "model.layers.9.mlp.experts.16.down_proj.weight", "model.layers.9.mlp.experts.17.gate_and_up_proj.weight", "model.layers.9.mlp.experts.17.down_proj.weight", "model.layers.9.mlp.experts.18.gate_and_up_proj.weight", "model.layers.9.mlp.experts.18.down_proj.weight", "model.layers.9.mlp.experts.19.gate_and_up_proj.weight", "model.layers.9.mlp.experts.19.down_proj.weight", "model.layers.9.mlp.experts.20.gate_and_up_proj.weight", "model.layers.9.mlp.experts.20.down_proj.weight", "model.layers.9.mlp.experts.21.gate_and_up_proj.weight", "model.layers.9.mlp.experts.21.down_proj.weight", "model.layers.9.mlp.experts.22.gate_and_up_proj.weight", "model.layers.9.mlp.experts.22.down_proj.weight", "model.layers.9.mlp.experts.23.gate_and_up_proj.weight", "model.layers.9.mlp.experts.23.down_proj.weight", "model.layers.9.mlp.experts.24.gate_and_up_proj.weight", "model.layers.9.mlp.experts.24.down_proj.weight", "model.layers.9.mlp.experts.25.gate_and_up_proj.weight", "model.layers.9.mlp.experts.25.down_proj.weight", "model.layers.9.mlp.experts.26.gate_and_up_proj.weight", "model.layers.9.mlp.experts.26.down_proj.weight", "model.layers.9.mlp.experts.27.gate_and_up_proj.weight", "model.layers.9.mlp.experts.27.down_proj.weight", "model.layers.9.mlp.experts.28.gate_and_up_proj.weight", "model.layers.9.mlp.experts.28.down_proj.weight", "model.layers.9.mlp.experts.29.gate_and_up_proj.weight", "model.layers.9.mlp.experts.29.down_proj.weight", "model.layers.9.mlp.experts.30.gate_and_up_proj.weight", "model.layers.9.mlp.experts.30.down_proj.weight", "model.layers.9.mlp.experts.31.gate_and_up_proj.weight", "model.layers.9.mlp.experts.31.down_proj.weight", "model.layers.9.mlp.experts.32.gate_and_up_proj.weight", "model.layers.9.mlp.experts.32.down_proj.weight", "model.layers.9.mlp.experts.33.gate_and_up_proj.weight", "model.layers.9.mlp.experts.33.down_proj.weight", "model.layers.9.mlp.experts.34.gate_and_up_proj.weight", "model.layers.9.mlp.experts.34.down_proj.weight", "model.layers.9.mlp.experts.35.gate_and_up_proj.weight", "model.layers.9.mlp.experts.35.down_proj.weight", "model.layers.9.mlp.experts.36.gate_and_up_proj.weight", "model.layers.9.mlp.experts.36.down_proj.weight", "model.layers.9.mlp.experts.37.gate_and_up_proj.weight", "model.layers.9.mlp.experts.37.down_proj.weight", "model.layers.9.mlp.experts.38.gate_and_up_proj.weight", "model.layers.9.mlp.experts.38.down_proj.weight", "model.layers.9.mlp.experts.39.gate_and_up_proj.weight", "model.layers.9.mlp.experts.39.down_proj.weight", "model.layers.9.mlp.experts.40.gate_and_up_proj.weight", "model.layers.9.mlp.experts.40.down_proj.weight", "model.layers.9.mlp.experts.41.gate_and_up_proj.weight", "model.layers.9.mlp.experts.41.down_proj.weight", "model.layers.9.mlp.experts.42.gate_and_up_proj.weight", "model.layers.9.mlp.experts.42.down_proj.weight", "model.layers.9.mlp.experts.43.gate_and_up_proj.weight", "model.layers.9.mlp.experts.43.down_proj.weight", "model.layers.9.mlp.experts.44.gate_and_up_proj.weight", "model.layers.9.mlp.experts.44.down_proj.weight", "model.layers.9.mlp.experts.45.gate_and_up_proj.weight", "model.layers.9.mlp.experts.45.down_proj.weight", "model.layers.9.mlp.experts.46.gate_and_up_proj.weight", "model.layers.9.mlp.experts.46.down_proj.weight", "model.layers.9.mlp.experts.47.gate_and_up_proj.weight", "model.layers.9.mlp.experts.47.down_proj.weight", "model.layers.9.mlp.experts.48.gate_and_up_proj.weight", "model.layers.9.mlp.experts.48.down_proj.weight", "model.layers.9.mlp.experts.49.gate_and_up_proj.weight", "model.layers.9.mlp.experts.49.down_proj.weight", "model.layers.9.mlp.experts.50.gate_and_up_proj.weight", "model.layers.9.mlp.experts.50.down_proj.weight", "model.layers.9.mlp.experts.51.gate_and_up_proj.weight", "model.layers.9.mlp.experts.51.down_proj.weight", "model.layers.9.mlp.experts.52.gate_and_up_proj.weight", "model.layers.9.mlp.experts.52.down_proj.weight", "model.layers.9.mlp.experts.53.gate_and_up_proj.weight", "model.layers.9.mlp.experts.53.down_proj.weight", "model.layers.9.mlp.experts.54.gate_and_up_proj.weight", "model.layers.9.mlp.experts.54.down_proj.weight", "model.layers.9.mlp.experts.55.gate_and_up_proj.weight", "model.layers.9.mlp.experts.55.down_proj.weight", "model.layers.9.mlp.experts.56.gate_and_up_proj.weight", "model.layers.9.mlp.experts.56.down_proj.weight", "model.layers.9.mlp.experts.57.gate_and_up_proj.weight", "model.layers.9.mlp.experts.57.down_proj.weight", "model.layers.9.mlp.experts.58.gate_and_up_proj.weight", "model.layers.9.mlp.experts.58.down_proj.weight", "model.layers.9.mlp.experts.59.gate_and_up_proj.weight", "model.layers.9.mlp.experts.59.down_proj.weight", "model.layers.9.mlp.experts.60.gate_and_up_proj.weight", "model.layers.9.mlp.experts.60.down_proj.weight", "model.layers.9.mlp.experts.61.gate_and_up_proj.weight", "model.layers.9.mlp.experts.61.down_proj.weight", "model.layers.9.mlp.experts.62.gate_and_up_proj.weight", "model.layers.9.mlp.experts.62.down_proj.weight", "model.layers.9.mlp.experts.63.gate_and_up_proj.weight", "model.layers.9.mlp.experts.63.down_proj.weight", "model.layers.10.self_attn.qkv_proj.weight", "model.layers.10.self_attn.o_proj.weight", "model.layers.10.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.10.mlp.shared_mlp.down_proj.weight", "model.layers.10.mlp.experts.0.gate_and_up_proj.weight", "model.layers.10.mlp.experts.0.down_proj.weight", "model.layers.10.mlp.experts.1.gate_and_up_proj.weight", "model.layers.10.mlp.experts.1.down_proj.weight", "model.layers.10.mlp.experts.2.gate_and_up_proj.weight", "model.layers.10.mlp.experts.2.down_proj.weight", "model.layers.10.mlp.experts.3.gate_and_up_proj.weight", "model.layers.10.mlp.experts.3.down_proj.weight", "model.layers.10.mlp.experts.4.gate_and_up_proj.weight", "model.layers.10.mlp.experts.4.down_proj.weight", "model.layers.10.mlp.experts.5.gate_and_up_proj.weight", "model.layers.10.mlp.experts.5.down_proj.weight", "model.layers.10.mlp.experts.6.gate_and_up_proj.weight", "model.layers.10.mlp.experts.6.down_proj.weight", "model.layers.10.mlp.experts.7.gate_and_up_proj.weight", "model.layers.10.mlp.experts.7.down_proj.weight", "model.layers.10.mlp.experts.8.gate_and_up_proj.weight", "model.layers.10.mlp.experts.8.down_proj.weight", "model.layers.10.mlp.experts.9.gate_and_up_proj.weight", "model.layers.10.mlp.experts.9.down_proj.weight", "model.layers.10.mlp.experts.10.gate_and_up_proj.weight", "model.layers.10.mlp.experts.10.down_proj.weight", "model.layers.10.mlp.experts.11.gate_and_up_proj.weight", "model.layers.10.mlp.experts.11.down_proj.weight", "model.layers.10.mlp.experts.12.gate_and_up_proj.weight", "model.layers.10.mlp.experts.12.down_proj.weight", "model.layers.10.mlp.experts.13.gate_and_up_proj.weight", "model.layers.10.mlp.experts.13.down_proj.weight", "model.layers.10.mlp.experts.14.gate_and_up_proj.weight", "model.layers.10.mlp.experts.14.down_proj.weight", "model.layers.10.mlp.experts.15.gate_and_up_proj.weight", "model.layers.10.mlp.experts.15.down_proj.weight", "model.layers.10.mlp.experts.16.gate_and_up_proj.weight", "model.layers.10.mlp.experts.16.down_proj.weight", "model.layers.10.mlp.experts.17.gate_and_up_proj.weight", "model.layers.10.mlp.experts.17.down_proj.weight", "model.layers.10.mlp.experts.18.gate_and_up_proj.weight", "model.layers.10.mlp.experts.18.down_proj.weight", "model.layers.10.mlp.experts.19.gate_and_up_proj.weight", "model.layers.10.mlp.experts.19.down_proj.weight", "model.layers.10.mlp.experts.20.gate_and_up_proj.weight", "model.layers.10.mlp.experts.20.down_proj.weight", "model.layers.10.mlp.experts.21.gate_and_up_proj.weight", "model.layers.10.mlp.experts.21.down_proj.weight", "model.layers.10.mlp.experts.22.gate_and_up_proj.weight", "model.layers.10.mlp.experts.22.down_proj.weight", "model.layers.10.mlp.experts.23.gate_and_up_proj.weight", "model.layers.10.mlp.experts.23.down_proj.weight", "model.layers.10.mlp.experts.24.gate_and_up_proj.weight", "model.layers.10.mlp.experts.24.down_proj.weight", "model.layers.10.mlp.experts.25.gate_and_up_proj.weight", "model.layers.10.mlp.experts.25.down_proj.weight", "model.layers.10.mlp.experts.26.gate_and_up_proj.weight", "model.layers.10.mlp.experts.26.down_proj.weight", "model.layers.10.mlp.experts.27.gate_and_up_proj.weight", "model.layers.10.mlp.experts.27.down_proj.weight", "model.layers.10.mlp.experts.28.gate_and_up_proj.weight", "model.layers.10.mlp.experts.28.down_proj.weight", "model.layers.10.mlp.experts.29.gate_and_up_proj.weight", "model.layers.10.mlp.experts.29.down_proj.weight", "model.layers.10.mlp.experts.30.gate_and_up_proj.weight", "model.layers.10.mlp.experts.30.down_proj.weight", "model.layers.10.mlp.experts.31.gate_and_up_proj.weight", "model.layers.10.mlp.experts.31.down_proj.weight", "model.layers.10.mlp.experts.32.gate_and_up_proj.weight", "model.layers.10.mlp.experts.32.down_proj.weight", "model.layers.10.mlp.experts.33.gate_and_up_proj.weight", "model.layers.10.mlp.experts.33.down_proj.weight", "model.layers.10.mlp.experts.34.gate_and_up_proj.weight", "model.layers.10.mlp.experts.34.down_proj.weight", "model.layers.10.mlp.experts.35.gate_and_up_proj.weight", "model.layers.10.mlp.experts.35.down_proj.weight", "model.layers.10.mlp.experts.36.gate_and_up_proj.weight", "model.layers.10.mlp.experts.36.down_proj.weight", "model.layers.10.mlp.experts.37.gate_and_up_proj.weight", "model.layers.10.mlp.experts.37.down_proj.weight", "model.layers.10.mlp.experts.38.gate_and_up_proj.weight", "model.layers.10.mlp.experts.38.down_proj.weight", "model.layers.10.mlp.experts.39.gate_and_up_proj.weight", "model.layers.10.mlp.experts.39.down_proj.weight", "model.layers.10.mlp.experts.40.gate_and_up_proj.weight", "model.layers.10.mlp.experts.40.down_proj.weight", "model.layers.10.mlp.experts.41.gate_and_up_proj.weight", "model.layers.10.mlp.experts.41.down_proj.weight", "model.layers.10.mlp.experts.42.gate_and_up_proj.weight", "model.layers.10.mlp.experts.42.down_proj.weight", "model.layers.10.mlp.experts.43.gate_and_up_proj.weight", "model.layers.10.mlp.experts.43.down_proj.weight", "model.layers.10.mlp.experts.44.gate_and_up_proj.weight", "model.layers.10.mlp.experts.44.down_proj.weight", "model.layers.10.mlp.experts.45.gate_and_up_proj.weight", "model.layers.10.mlp.experts.45.down_proj.weight", "model.layers.10.mlp.experts.46.gate_and_up_proj.weight", "model.layers.10.mlp.experts.46.down_proj.weight", "model.layers.10.mlp.experts.47.gate_and_up_proj.weight", "model.layers.10.mlp.experts.47.down_proj.weight", "model.layers.10.mlp.experts.48.gate_and_up_proj.weight", "model.layers.10.mlp.experts.48.down_proj.weight", "model.layers.10.mlp.experts.49.gate_and_up_proj.weight", "model.layers.10.mlp.experts.49.down_proj.weight", "model.layers.10.mlp.experts.50.gate_and_up_proj.weight", "model.layers.10.mlp.experts.50.down_proj.weight", "model.layers.10.mlp.experts.51.gate_and_up_proj.weight", "model.layers.10.mlp.experts.51.down_proj.weight", "model.layers.10.mlp.experts.52.gate_and_up_proj.weight", "model.layers.10.mlp.experts.52.down_proj.weight", "model.layers.10.mlp.experts.53.gate_and_up_proj.weight", "model.layers.10.mlp.experts.53.down_proj.weight", "model.layers.10.mlp.experts.54.gate_and_up_proj.weight", "model.layers.10.mlp.experts.54.down_proj.weight", "model.layers.10.mlp.experts.55.gate_and_up_proj.weight", "model.layers.10.mlp.experts.55.down_proj.weight", "model.layers.10.mlp.experts.56.gate_and_up_proj.weight", "model.layers.10.mlp.experts.56.down_proj.weight", "model.layers.10.mlp.experts.57.gate_and_up_proj.weight", "model.layers.10.mlp.experts.57.down_proj.weight", "model.layers.10.mlp.experts.58.gate_and_up_proj.weight", "model.layers.10.mlp.experts.58.down_proj.weight", "model.layers.10.mlp.experts.59.gate_and_up_proj.weight", "model.layers.10.mlp.experts.59.down_proj.weight", "model.layers.10.mlp.experts.60.gate_and_up_proj.weight", "model.layers.10.mlp.experts.60.down_proj.weight", "model.layers.10.mlp.experts.61.gate_and_up_proj.weight", "model.layers.10.mlp.experts.61.down_proj.weight", "model.layers.10.mlp.experts.62.gate_and_up_proj.weight", "model.layers.10.mlp.experts.62.down_proj.weight", "model.layers.10.mlp.experts.63.gate_and_up_proj.weight", "model.layers.10.mlp.experts.63.down_proj.weight", "model.layers.11.self_attn.qkv_proj.weight", "model.layers.11.self_attn.o_proj.weight", "model.layers.11.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.11.mlp.shared_mlp.down_proj.weight", "model.layers.11.mlp.experts.0.gate_and_up_proj.weight", "model.layers.11.mlp.experts.0.down_proj.weight", "model.layers.11.mlp.experts.1.gate_and_up_proj.weight", "model.layers.11.mlp.experts.1.down_proj.weight", "model.layers.11.mlp.experts.2.gate_and_up_proj.weight", "model.layers.11.mlp.experts.2.down_proj.weight", "model.layers.11.mlp.experts.3.gate_and_up_proj.weight", "model.layers.11.mlp.experts.3.down_proj.weight", "model.layers.11.mlp.experts.4.gate_and_up_proj.weight", "model.layers.11.mlp.experts.4.down_proj.weight", "model.layers.11.mlp.experts.5.gate_and_up_proj.weight", "model.layers.11.mlp.experts.5.down_proj.weight", "model.layers.11.mlp.experts.6.gate_and_up_proj.weight", "model.layers.11.mlp.experts.6.down_proj.weight", "model.layers.11.mlp.experts.7.gate_and_up_proj.weight", "model.layers.11.mlp.experts.7.down_proj.weight", "model.layers.11.mlp.experts.8.gate_and_up_proj.weight", "model.layers.11.mlp.experts.8.down_proj.weight", "model.layers.11.mlp.experts.9.gate_and_up_proj.weight", "model.layers.11.mlp.experts.9.down_proj.weight", "model.layers.11.mlp.experts.10.gate_and_up_proj.weight", "model.layers.11.mlp.experts.10.down_proj.weight", "model.layers.11.mlp.experts.11.gate_and_up_proj.weight", "model.layers.11.mlp.experts.11.down_proj.weight", "model.layers.11.mlp.experts.12.gate_and_up_proj.weight", "model.layers.11.mlp.experts.12.down_proj.weight", "model.layers.11.mlp.experts.13.gate_and_up_proj.weight", "model.layers.11.mlp.experts.13.down_proj.weight", "model.layers.11.mlp.experts.14.gate_and_up_proj.weight", "model.layers.11.mlp.experts.14.down_proj.weight", "model.layers.11.mlp.experts.15.gate_and_up_proj.weight", "model.layers.11.mlp.experts.15.down_proj.weight", "model.layers.11.mlp.experts.16.gate_and_up_proj.weight", "model.layers.11.mlp.experts.16.down_proj.weight", "model.layers.11.mlp.experts.17.gate_and_up_proj.weight", "model.layers.11.mlp.experts.17.down_proj.weight", "model.layers.11.mlp.experts.18.gate_and_up_proj.weight", "model.layers.11.mlp.experts.18.down_proj.weight", "model.layers.11.mlp.experts.19.gate_and_up_proj.weight", "model.layers.11.mlp.experts.19.down_proj.weight", "model.layers.11.mlp.experts.20.gate_and_up_proj.weight", "model.layers.11.mlp.experts.20.down_proj.weight", "model.layers.11.mlp.experts.21.gate_and_up_proj.weight", "model.layers.11.mlp.experts.21.down_proj.weight", "model.layers.11.mlp.experts.22.gate_and_up_proj.weight", "model.layers.11.mlp.experts.22.down_proj.weight", "model.layers.11.mlp.experts.23.gate_and_up_proj.weight", "model.layers.11.mlp.experts.23.down_proj.weight", "model.layers.11.mlp.experts.24.gate_and_up_proj.weight", "model.layers.11.mlp.experts.24.down_proj.weight", "model.layers.11.mlp.experts.25.gate_and_up_proj.weight", "model.layers.11.mlp.experts.25.down_proj.weight", "model.layers.11.mlp.experts.26.gate_and_up_proj.weight", "model.layers.11.mlp.experts.26.down_proj.weight", "model.layers.11.mlp.experts.27.gate_and_up_proj.weight", "model.layers.11.mlp.experts.27.down_proj.weight", "model.layers.11.mlp.experts.28.gate_and_up_proj.weight", "model.layers.11.mlp.experts.28.down_proj.weight", "model.layers.11.mlp.experts.29.gate_and_up_proj.weight", "model.layers.11.mlp.experts.29.down_proj.weight", "model.layers.11.mlp.experts.30.gate_and_up_proj.weight", "model.layers.11.mlp.experts.30.down_proj.weight", "model.layers.11.mlp.experts.31.gate_and_up_proj.weight", "model.layers.11.mlp.experts.31.down_proj.weight", "model.layers.11.mlp.experts.32.gate_and_up_proj.weight", "model.layers.11.mlp.experts.32.down_proj.weight", "model.layers.11.mlp.experts.33.gate_and_up_proj.weight", "model.layers.11.mlp.experts.33.down_proj.weight", "model.layers.11.mlp.experts.34.gate_and_up_proj.weight", "model.layers.11.mlp.experts.34.down_proj.weight", "model.layers.11.mlp.experts.35.gate_and_up_proj.weight", "model.layers.11.mlp.experts.35.down_proj.weight", "model.layers.11.mlp.experts.36.gate_and_up_proj.weight", "model.layers.11.mlp.experts.36.down_proj.weight", "model.layers.11.mlp.experts.37.gate_and_up_proj.weight", "model.layers.11.mlp.experts.37.down_proj.weight", "model.layers.11.mlp.experts.38.gate_and_up_proj.weight", "model.layers.11.mlp.experts.38.down_proj.weight", "model.layers.11.mlp.experts.39.gate_and_up_proj.weight", "model.layers.11.mlp.experts.39.down_proj.weight", "model.layers.11.mlp.experts.40.gate_and_up_proj.weight", "model.layers.11.mlp.experts.40.down_proj.weight", "model.layers.11.mlp.experts.41.gate_and_up_proj.weight", "model.layers.11.mlp.experts.41.down_proj.weight", "model.layers.11.mlp.experts.42.gate_and_up_proj.weight", "model.layers.11.mlp.experts.42.down_proj.weight", "model.layers.11.mlp.experts.43.gate_and_up_proj.weight", "model.layers.11.mlp.experts.43.down_proj.weight", "model.layers.11.mlp.experts.44.gate_and_up_proj.weight", "model.layers.11.mlp.experts.44.down_proj.weight", "model.layers.11.mlp.experts.45.gate_and_up_proj.weight", "model.layers.11.mlp.experts.45.down_proj.weight", "model.layers.11.mlp.experts.46.gate_and_up_proj.weight", "model.layers.11.mlp.experts.46.down_proj.weight", "model.layers.11.mlp.experts.47.gate_and_up_proj.weight", "model.layers.11.mlp.experts.47.down_proj.weight", "model.layers.11.mlp.experts.48.gate_and_up_proj.weight", "model.layers.11.mlp.experts.48.down_proj.weight", "model.layers.11.mlp.experts.49.gate_and_up_proj.weight", "model.layers.11.mlp.experts.49.down_proj.weight", "model.layers.11.mlp.experts.50.gate_and_up_proj.weight", "model.layers.11.mlp.experts.50.down_proj.weight", "model.layers.11.mlp.experts.51.gate_and_up_proj.weight", "model.layers.11.mlp.experts.51.down_proj.weight", "model.layers.11.mlp.experts.52.gate_and_up_proj.weight", "model.layers.11.mlp.experts.52.down_proj.weight", "model.layers.11.mlp.experts.53.gate_and_up_proj.weight", "model.layers.11.mlp.experts.53.down_proj.weight", "model.layers.11.mlp.experts.54.gate_and_up_proj.weight", "model.layers.11.mlp.experts.54.down_proj.weight", "model.layers.11.mlp.experts.55.gate_and_up_proj.weight", "model.layers.11.mlp.experts.55.down_proj.weight", "model.layers.11.mlp.experts.56.gate_and_up_proj.weight", "model.layers.11.mlp.experts.56.down_proj.weight", "model.layers.11.mlp.experts.57.gate_and_up_proj.weight", "model.layers.11.mlp.experts.57.down_proj.weight", "model.layers.11.mlp.experts.58.gate_and_up_proj.weight", "model.layers.11.mlp.experts.58.down_proj.weight", "model.layers.11.mlp.experts.59.gate_and_up_proj.weight", "model.layers.11.mlp.experts.59.down_proj.weight", "model.layers.11.mlp.experts.60.gate_and_up_proj.weight", "model.layers.11.mlp.experts.60.down_proj.weight", "model.layers.11.mlp.experts.61.gate_and_up_proj.weight", "model.layers.11.mlp.experts.61.down_proj.weight", "model.layers.11.mlp.experts.62.gate_and_up_proj.weight", "model.layers.11.mlp.experts.62.down_proj.weight", "model.layers.11.mlp.experts.63.gate_and_up_proj.weight", "model.layers.11.mlp.experts.63.down_proj.weight", "model.layers.12.self_attn.qkv_proj.weight", "model.layers.12.self_attn.o_proj.weight", "model.layers.12.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.12.mlp.shared_mlp.down_proj.weight", "model.layers.12.mlp.experts.0.gate_and_up_proj.weight", "model.layers.12.mlp.experts.0.down_proj.weight", "model.layers.12.mlp.experts.1.gate_and_up_proj.weight", "model.layers.12.mlp.experts.1.down_proj.weight", "model.layers.12.mlp.experts.2.gate_and_up_proj.weight", "model.layers.12.mlp.experts.2.down_proj.weight", "model.layers.12.mlp.experts.3.gate_and_up_proj.weight", "model.layers.12.mlp.experts.3.down_proj.weight", "model.layers.12.mlp.experts.4.gate_and_up_proj.weight", "model.layers.12.mlp.experts.4.down_proj.weight", "model.layers.12.mlp.experts.5.gate_and_up_proj.weight", "model.layers.12.mlp.experts.5.down_proj.weight", "model.layers.12.mlp.experts.6.gate_and_up_proj.weight", "model.layers.12.mlp.experts.6.down_proj.weight", "model.layers.12.mlp.experts.7.gate_and_up_proj.weight", "model.layers.12.mlp.experts.7.down_proj.weight", "model.layers.12.mlp.experts.8.gate_and_up_proj.weight", "model.layers.12.mlp.experts.8.down_proj.weight", "model.layers.12.mlp.experts.9.gate_and_up_proj.weight", "model.layers.12.mlp.experts.9.down_proj.weight", "model.layers.12.mlp.experts.10.gate_and_up_proj.weight", "model.layers.12.mlp.experts.10.down_proj.weight", "model.layers.12.mlp.experts.11.gate_and_up_proj.weight", "model.layers.12.mlp.experts.11.down_proj.weight", "model.layers.12.mlp.experts.12.gate_and_up_proj.weight", "model.layers.12.mlp.experts.12.down_proj.weight", "model.layers.12.mlp.experts.13.gate_and_up_proj.weight", "model.layers.12.mlp.experts.13.down_proj.weight", "model.layers.12.mlp.experts.14.gate_and_up_proj.weight", "model.layers.12.mlp.experts.14.down_proj.weight", "model.layers.12.mlp.experts.15.gate_and_up_proj.weight", "model.layers.12.mlp.experts.15.down_proj.weight", "model.layers.12.mlp.experts.16.gate_and_up_proj.weight", "model.layers.12.mlp.experts.16.down_proj.weight", "model.layers.12.mlp.experts.17.gate_and_up_proj.weight", "model.layers.12.mlp.experts.17.down_proj.weight", "model.layers.12.mlp.experts.18.gate_and_up_proj.weight", "model.layers.12.mlp.experts.18.down_proj.weight", "model.layers.12.mlp.experts.19.gate_and_up_proj.weight", "model.layers.12.mlp.experts.19.down_proj.weight", "model.layers.12.mlp.experts.20.gate_and_up_proj.weight", "model.layers.12.mlp.experts.20.down_proj.weight", "model.layers.12.mlp.experts.21.gate_and_up_proj.weight", "model.layers.12.mlp.experts.21.down_proj.weight", "model.layers.12.mlp.experts.22.gate_and_up_proj.weight", "model.layers.12.mlp.experts.22.down_proj.weight", "model.layers.12.mlp.experts.23.gate_and_up_proj.weight", "model.layers.12.mlp.experts.23.down_proj.weight", "model.layers.12.mlp.experts.24.gate_and_up_proj.weight", "model.layers.12.mlp.experts.24.down_proj.weight", "model.layers.12.mlp.experts.25.gate_and_up_proj.weight", "model.layers.12.mlp.experts.25.down_proj.weight", "model.layers.12.mlp.experts.26.gate_and_up_proj.weight", "model.layers.12.mlp.experts.26.down_proj.weight", "model.layers.12.mlp.experts.27.gate_and_up_proj.weight", "model.layers.12.mlp.experts.27.down_proj.weight", "model.layers.12.mlp.experts.28.gate_and_up_proj.weight", "model.layers.12.mlp.experts.28.down_proj.weight", "model.layers.12.mlp.experts.29.gate_and_up_proj.weight", "model.layers.12.mlp.experts.29.down_proj.weight", "model.layers.12.mlp.experts.30.gate_and_up_proj.weight", "model.layers.12.mlp.experts.30.down_proj.weight", "model.layers.12.mlp.experts.31.gate_and_up_proj.weight", "model.layers.12.mlp.experts.31.down_proj.weight", "model.layers.12.mlp.experts.32.gate_and_up_proj.weight", "model.layers.12.mlp.experts.32.down_proj.weight", "model.layers.12.mlp.experts.33.gate_and_up_proj.weight", "model.layers.12.mlp.experts.33.down_proj.weight", "model.layers.12.mlp.experts.34.gate_and_up_proj.weight", "model.layers.12.mlp.experts.34.down_proj.weight", "model.layers.12.mlp.experts.35.gate_and_up_proj.weight", "model.layers.12.mlp.experts.35.down_proj.weight", "model.layers.12.mlp.experts.36.gate_and_up_proj.weight", "model.layers.12.mlp.experts.36.down_proj.weight", "model.layers.12.mlp.experts.37.gate_and_up_proj.weight", "model.layers.12.mlp.experts.37.down_proj.weight", "model.layers.12.mlp.experts.38.gate_and_up_proj.weight", "model.layers.12.mlp.experts.38.down_proj.weight", "model.layers.12.mlp.experts.39.gate_and_up_proj.weight", "model.layers.12.mlp.experts.39.down_proj.weight", "model.layers.12.mlp.experts.40.gate_and_up_proj.weight", "model.layers.12.mlp.experts.40.down_proj.weight", "model.layers.12.mlp.experts.41.gate_and_up_proj.weight", "model.layers.12.mlp.experts.41.down_proj.weight", "model.layers.12.mlp.experts.42.gate_and_up_proj.weight", "model.layers.12.mlp.experts.42.down_proj.weight", "model.layers.12.mlp.experts.43.gate_and_up_proj.weight", "model.layers.12.mlp.experts.43.down_proj.weight", "model.layers.12.mlp.experts.44.gate_and_up_proj.weight", "model.layers.12.mlp.experts.44.down_proj.weight", "model.layers.12.mlp.experts.45.gate_and_up_proj.weight", "model.layers.12.mlp.experts.45.down_proj.weight", "model.layers.12.mlp.experts.46.gate_and_up_proj.weight", "model.layers.12.mlp.experts.46.down_proj.weight", "model.layers.12.mlp.experts.47.gate_and_up_proj.weight", "model.layers.12.mlp.experts.47.down_proj.weight", "model.layers.12.mlp.experts.48.gate_and_up_proj.weight", "model.layers.12.mlp.experts.48.down_proj.weight", "model.layers.12.mlp.experts.49.gate_and_up_proj.weight", "model.layers.12.mlp.experts.49.down_proj.weight", "model.layers.12.mlp.experts.50.gate_and_up_proj.weight", "model.layers.12.mlp.experts.50.down_proj.weight", "model.layers.12.mlp.experts.51.gate_and_up_proj.weight", "model.layers.12.mlp.experts.51.down_proj.weight", "model.layers.12.mlp.experts.52.gate_and_up_proj.weight", "model.layers.12.mlp.experts.52.down_proj.weight", "model.layers.12.mlp.experts.53.gate_and_up_proj.weight", "model.layers.12.mlp.experts.53.down_proj.weight", "model.layers.12.mlp.experts.54.gate_and_up_proj.weight", "model.layers.12.mlp.experts.54.down_proj.weight", "model.layers.12.mlp.experts.55.gate_and_up_proj.weight", "model.layers.12.mlp.experts.55.down_proj.weight", "model.layers.12.mlp.experts.56.gate_and_up_proj.weight", "model.layers.12.mlp.experts.56.down_proj.weight", "model.layers.12.mlp.experts.57.gate_and_up_proj.weight", "model.layers.12.mlp.experts.57.down_proj.weight", "model.layers.12.mlp.experts.58.gate_and_up_proj.weight", "model.layers.12.mlp.experts.58.down_proj.weight", "model.layers.12.mlp.experts.59.gate_and_up_proj.weight", "model.layers.12.mlp.experts.59.down_proj.weight", "model.layers.12.mlp.experts.60.gate_and_up_proj.weight", "model.layers.12.mlp.experts.60.down_proj.weight", "model.layers.12.mlp.experts.61.gate_and_up_proj.weight", "model.layers.12.mlp.experts.61.down_proj.weight", "model.layers.12.mlp.experts.62.gate_and_up_proj.weight", "model.layers.12.mlp.experts.62.down_proj.weight", "model.layers.12.mlp.experts.63.gate_and_up_proj.weight", "model.layers.12.mlp.experts.63.down_proj.weight", "model.layers.13.self_attn.qkv_proj.weight", "model.layers.13.self_attn.o_proj.weight", "model.layers.13.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.13.mlp.shared_mlp.down_proj.weight", "model.layers.13.mlp.experts.0.gate_and_up_proj.weight", "model.layers.13.mlp.experts.0.down_proj.weight", "model.layers.13.mlp.experts.1.gate_and_up_proj.weight", "model.layers.13.mlp.experts.1.down_proj.weight", "model.layers.13.mlp.experts.2.gate_and_up_proj.weight", "model.layers.13.mlp.experts.2.down_proj.weight", "model.layers.13.mlp.experts.3.gate_and_up_proj.weight", "model.layers.13.mlp.experts.3.down_proj.weight", "model.layers.13.mlp.experts.4.gate_and_up_proj.weight", "model.layers.13.mlp.experts.4.down_proj.weight", "model.layers.13.mlp.experts.5.gate_and_up_proj.weight", "model.layers.13.mlp.experts.5.down_proj.weight", "model.layers.13.mlp.experts.6.gate_and_up_proj.weight", "model.layers.13.mlp.experts.6.down_proj.weight", "model.layers.13.mlp.experts.7.gate_and_up_proj.weight", "model.layers.13.mlp.experts.7.down_proj.weight", "model.layers.13.mlp.experts.8.gate_and_up_proj.weight", "model.layers.13.mlp.experts.8.down_proj.weight", "model.layers.13.mlp.experts.9.gate_and_up_proj.weight", "model.layers.13.mlp.experts.9.down_proj.weight", "model.layers.13.mlp.experts.10.gate_and_up_proj.weight", "model.layers.13.mlp.experts.10.down_proj.weight", "model.layers.13.mlp.experts.11.gate_and_up_proj.weight", "model.layers.13.mlp.experts.11.down_proj.weight", "model.layers.13.mlp.experts.12.gate_and_up_proj.weight", "model.layers.13.mlp.experts.12.down_proj.weight", "model.layers.13.mlp.experts.13.gate_and_up_proj.weight", "model.layers.13.mlp.experts.13.down_proj.weight", "model.layers.13.mlp.experts.14.gate_and_up_proj.weight", "model.layers.13.mlp.experts.14.down_proj.weight", "model.layers.13.mlp.experts.15.gate_and_up_proj.weight", "model.layers.13.mlp.experts.15.down_proj.weight", "model.layers.13.mlp.experts.16.gate_and_up_proj.weight", "model.layers.13.mlp.experts.16.down_proj.weight", "model.layers.13.mlp.experts.17.gate_and_up_proj.weight", "model.layers.13.mlp.experts.17.down_proj.weight", "model.layers.13.mlp.experts.18.gate_and_up_proj.weight", "model.layers.13.mlp.experts.18.down_proj.weight", "model.layers.13.mlp.experts.19.gate_and_up_proj.weight", "model.layers.13.mlp.experts.19.down_proj.weight", "model.layers.13.mlp.experts.20.gate_and_up_proj.weight", "model.layers.13.mlp.experts.20.down_proj.weight", "model.layers.13.mlp.experts.21.gate_and_up_proj.weight", "model.layers.13.mlp.experts.21.down_proj.weight", "model.layers.13.mlp.experts.22.gate_and_up_proj.weight", "model.layers.13.mlp.experts.22.down_proj.weight", "model.layers.13.mlp.experts.23.gate_and_up_proj.weight", "model.layers.13.mlp.experts.23.down_proj.weight", "model.layers.13.mlp.experts.24.gate_and_up_proj.weight", "model.layers.13.mlp.experts.24.down_proj.weight", "model.layers.13.mlp.experts.25.gate_and_up_proj.weight", "model.layers.13.mlp.experts.25.down_proj.weight", "model.layers.13.mlp.experts.26.gate_and_up_proj.weight", "model.layers.13.mlp.experts.26.down_proj.weight", "model.layers.13.mlp.experts.27.gate_and_up_proj.weight", "model.layers.13.mlp.experts.27.down_proj.weight", "model.layers.13.mlp.experts.28.gate_and_up_proj.weight", "model.layers.13.mlp.experts.28.down_proj.weight", "model.layers.13.mlp.experts.29.gate_and_up_proj.weight", "model.layers.13.mlp.experts.29.down_proj.weight", "model.layers.13.mlp.experts.30.gate_and_up_proj.weight", "model.layers.13.mlp.experts.30.down_proj.weight", "model.layers.13.mlp.experts.31.gate_and_up_proj.weight", "model.layers.13.mlp.experts.31.down_proj.weight", "model.layers.13.mlp.experts.32.gate_and_up_proj.weight", "model.layers.13.mlp.experts.32.down_proj.weight", "model.layers.13.mlp.experts.33.gate_and_up_proj.weight", "model.layers.13.mlp.experts.33.down_proj.weight", "model.layers.13.mlp.experts.34.gate_and_up_proj.weight", "model.layers.13.mlp.experts.34.down_proj.weight", "model.layers.13.mlp.experts.35.gate_and_up_proj.weight", "model.layers.13.mlp.experts.35.down_proj.weight", "model.layers.13.mlp.experts.36.gate_and_up_proj.weight", "model.layers.13.mlp.experts.36.down_proj.weight", "model.layers.13.mlp.experts.37.gate_and_up_proj.weight", "model.layers.13.mlp.experts.37.down_proj.weight", "model.layers.13.mlp.experts.38.gate_and_up_proj.weight", "model.layers.13.mlp.experts.38.down_proj.weight", "model.layers.13.mlp.experts.39.gate_and_up_proj.weight", "model.layers.13.mlp.experts.39.down_proj.weight", "model.layers.13.mlp.experts.40.gate_and_up_proj.weight", "model.layers.13.mlp.experts.40.down_proj.weight", "model.layers.13.mlp.experts.41.gate_and_up_proj.weight", "model.layers.13.mlp.experts.41.down_proj.weight", "model.layers.13.mlp.experts.42.gate_and_up_proj.weight", "model.layers.13.mlp.experts.42.down_proj.weight", "model.layers.13.mlp.experts.43.gate_and_up_proj.weight", "model.layers.13.mlp.experts.43.down_proj.weight", "model.layers.13.mlp.experts.44.gate_and_up_proj.weight", "model.layers.13.mlp.experts.44.down_proj.weight", "model.layers.13.mlp.experts.45.gate_and_up_proj.weight", "model.layers.13.mlp.experts.45.down_proj.weight", "model.layers.13.mlp.experts.46.gate_and_up_proj.weight", "model.layers.13.mlp.experts.46.down_proj.weight", "model.layers.13.mlp.experts.47.gate_and_up_proj.weight", "model.layers.13.mlp.experts.47.down_proj.weight", "model.layers.13.mlp.experts.48.gate_and_up_proj.weight", "model.layers.13.mlp.experts.48.down_proj.weight", "model.layers.13.mlp.experts.49.gate_and_up_proj.weight", "model.layers.13.mlp.experts.49.down_proj.weight", "model.layers.13.mlp.experts.50.gate_and_up_proj.weight", "model.layers.13.mlp.experts.50.down_proj.weight", "model.layers.13.mlp.experts.51.gate_and_up_proj.weight", "model.layers.13.mlp.experts.51.down_proj.weight", "model.layers.13.mlp.experts.52.gate_and_up_proj.weight", "model.layers.13.mlp.experts.52.down_proj.weight", "model.layers.13.mlp.experts.53.gate_and_up_proj.weight", "model.layers.13.mlp.experts.53.down_proj.weight", "model.layers.13.mlp.experts.54.gate_and_up_proj.weight", "model.layers.13.mlp.experts.54.down_proj.weight", "model.layers.13.mlp.experts.55.gate_and_up_proj.weight", "model.layers.13.mlp.experts.55.down_proj.weight", "model.layers.13.mlp.experts.56.gate_and_up_proj.weight", "model.layers.13.mlp.experts.56.down_proj.weight", "model.layers.13.mlp.experts.57.gate_and_up_proj.weight", "model.layers.13.mlp.experts.57.down_proj.weight", "model.layers.13.mlp.experts.58.gate_and_up_proj.weight", "model.layers.13.mlp.experts.58.down_proj.weight", "model.layers.13.mlp.experts.59.gate_and_up_proj.weight", "model.layers.13.mlp.experts.59.down_proj.weight", "model.layers.13.mlp.experts.60.gate_and_up_proj.weight", "model.layers.13.mlp.experts.60.down_proj.weight", "model.layers.13.mlp.experts.61.gate_and_up_proj.weight", "model.layers.13.mlp.experts.61.down_proj.weight", "model.layers.13.mlp.experts.62.gate_and_up_proj.weight", "model.layers.13.mlp.experts.62.down_proj.weight", "model.layers.13.mlp.experts.63.gate_and_up_proj.weight", "model.layers.13.mlp.experts.63.down_proj.weight", "model.layers.14.self_attn.qkv_proj.weight", "model.layers.14.self_attn.o_proj.weight", "model.layers.14.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.14.mlp.shared_mlp.down_proj.weight", "model.layers.14.mlp.experts.0.gate_and_up_proj.weight", "model.layers.14.mlp.experts.0.down_proj.weight", "model.layers.14.mlp.experts.1.gate_and_up_proj.weight", "model.layers.14.mlp.experts.1.down_proj.weight", "model.layers.14.mlp.experts.2.gate_and_up_proj.weight", "model.layers.14.mlp.experts.2.down_proj.weight", "model.layers.14.mlp.experts.3.gate_and_up_proj.weight", "model.layers.14.mlp.experts.3.down_proj.weight", "model.layers.14.mlp.experts.4.gate_and_up_proj.weight", "model.layers.14.mlp.experts.4.down_proj.weight", "model.layers.14.mlp.experts.5.gate_and_up_proj.weight", "model.layers.14.mlp.experts.5.down_proj.weight", "model.layers.14.mlp.experts.6.gate_and_up_proj.weight", "model.layers.14.mlp.experts.6.down_proj.weight", "model.layers.14.mlp.experts.7.gate_and_up_proj.weight", "model.layers.14.mlp.experts.7.down_proj.weight", "model.layers.14.mlp.experts.8.gate_and_up_proj.weight", "model.layers.14.mlp.experts.8.down_proj.weight", "model.layers.14.mlp.experts.9.gate_and_up_proj.weight", "model.layers.14.mlp.experts.9.down_proj.weight", "model.layers.14.mlp.experts.10.gate_and_up_proj.weight", "model.layers.14.mlp.experts.10.down_proj.weight", "model.layers.14.mlp.experts.11.gate_and_up_proj.weight", "model.layers.14.mlp.experts.11.down_proj.weight", "model.layers.14.mlp.experts.12.gate_and_up_proj.weight", "model.layers.14.mlp.experts.12.down_proj.weight", "model.layers.14.mlp.experts.13.gate_and_up_proj.weight", "model.layers.14.mlp.experts.13.down_proj.weight", "model.layers.14.mlp.experts.14.gate_and_up_proj.weight", "model.layers.14.mlp.experts.14.down_proj.weight", "model.layers.14.mlp.experts.15.gate_and_up_proj.weight", "model.layers.14.mlp.experts.15.down_proj.weight", "model.layers.14.mlp.experts.16.gate_and_up_proj.weight", "model.layers.14.mlp.experts.16.down_proj.weight", "model.layers.14.mlp.experts.17.gate_and_up_proj.weight", "model.layers.14.mlp.experts.17.down_proj.weight", "model.layers.14.mlp.experts.18.gate_and_up_proj.weight", "model.layers.14.mlp.experts.18.down_proj.weight", "model.layers.14.mlp.experts.19.gate_and_up_proj.weight", "model.layers.14.mlp.experts.19.down_proj.weight", "model.layers.14.mlp.experts.20.gate_and_up_proj.weight", "model.layers.14.mlp.experts.20.down_proj.weight", "model.layers.14.mlp.experts.21.gate_and_up_proj.weight", "model.layers.14.mlp.experts.21.down_proj.weight", "model.layers.14.mlp.experts.22.gate_and_up_proj.weight", "model.layers.14.mlp.experts.22.down_proj.weight", "model.layers.14.mlp.experts.23.gate_and_up_proj.weight", "model.layers.14.mlp.experts.23.down_proj.weight", "model.layers.14.mlp.experts.24.gate_and_up_proj.weight", "model.layers.14.mlp.experts.24.down_proj.weight", "model.layers.14.mlp.experts.25.gate_and_up_proj.weight", "model.layers.14.mlp.experts.25.down_proj.weight", "model.layers.14.mlp.experts.26.gate_and_up_proj.weight", "model.layers.14.mlp.experts.26.down_proj.weight", "model.layers.14.mlp.experts.27.gate_and_up_proj.weight", "model.layers.14.mlp.experts.27.down_proj.weight", "model.layers.14.mlp.experts.28.gate_and_up_proj.weight", "model.layers.14.mlp.experts.28.down_proj.weight", "model.layers.14.mlp.experts.29.gate_and_up_proj.weight", "model.layers.14.mlp.experts.29.down_proj.weight", "model.layers.14.mlp.experts.30.gate_and_up_proj.weight", "model.layers.14.mlp.experts.30.down_proj.weight", "model.layers.14.mlp.experts.31.gate_and_up_proj.weight", "model.layers.14.mlp.experts.31.down_proj.weight", "model.layers.14.mlp.experts.32.gate_and_up_proj.weight", "model.layers.14.mlp.experts.32.down_proj.weight", "model.layers.14.mlp.experts.33.gate_and_up_proj.weight", "model.layers.14.mlp.experts.33.down_proj.weight", "model.layers.14.mlp.experts.34.gate_and_up_proj.weight", "model.layers.14.mlp.experts.34.down_proj.weight", "model.layers.14.mlp.experts.35.gate_and_up_proj.weight", "model.layers.14.mlp.experts.35.down_proj.weight", "model.layers.14.mlp.experts.36.gate_and_up_proj.weight", "model.layers.14.mlp.experts.36.down_proj.weight", "model.layers.14.mlp.experts.37.gate_and_up_proj.weight", "model.layers.14.mlp.experts.37.down_proj.weight", "model.layers.14.mlp.experts.38.gate_and_up_proj.weight", "model.layers.14.mlp.experts.38.down_proj.weight", "model.layers.14.mlp.experts.39.gate_and_up_proj.weight", "model.layers.14.mlp.experts.39.down_proj.weight", "model.layers.14.mlp.experts.40.gate_and_up_proj.weight", "model.layers.14.mlp.experts.40.down_proj.weight", "model.layers.14.mlp.experts.41.gate_and_up_proj.weight", "model.layers.14.mlp.experts.41.down_proj.weight", "model.layers.14.mlp.experts.42.gate_and_up_proj.weight", "model.layers.14.mlp.experts.42.down_proj.weight", "model.layers.14.mlp.experts.43.gate_and_up_proj.weight", "model.layers.14.mlp.experts.43.down_proj.weight", "model.layers.14.mlp.experts.44.gate_and_up_proj.weight", "model.layers.14.mlp.experts.44.down_proj.weight", "model.layers.14.mlp.experts.45.gate_and_up_proj.weight", "model.layers.14.mlp.experts.45.down_proj.weight", "model.layers.14.mlp.experts.46.gate_and_up_proj.weight", "model.layers.14.mlp.experts.46.down_proj.weight", "model.layers.14.mlp.experts.47.gate_and_up_proj.weight", "model.layers.14.mlp.experts.47.down_proj.weight", "model.layers.14.mlp.experts.48.gate_and_up_proj.weight", "model.layers.14.mlp.experts.48.down_proj.weight", "model.layers.14.mlp.experts.49.gate_and_up_proj.weight", "model.layers.14.mlp.experts.49.down_proj.weight", "model.layers.14.mlp.experts.50.gate_and_up_proj.weight", "model.layers.14.mlp.experts.50.down_proj.weight", "model.layers.14.mlp.experts.51.gate_and_up_proj.weight", "model.layers.14.mlp.experts.51.down_proj.weight", "model.layers.14.mlp.experts.52.gate_and_up_proj.weight", "model.layers.14.mlp.experts.52.down_proj.weight", "model.layers.14.mlp.experts.53.gate_and_up_proj.weight", "model.layers.14.mlp.experts.53.down_proj.weight", "model.layers.14.mlp.experts.54.gate_and_up_proj.weight", "model.layers.14.mlp.experts.54.down_proj.weight", "model.layers.14.mlp.experts.55.gate_and_up_proj.weight", "model.layers.14.mlp.experts.55.down_proj.weight", "model.layers.14.mlp.experts.56.gate_and_up_proj.weight", "model.layers.14.mlp.experts.56.down_proj.weight", "model.layers.14.mlp.experts.57.gate_and_up_proj.weight", "model.layers.14.mlp.experts.57.down_proj.weight", "model.layers.14.mlp.experts.58.gate_and_up_proj.weight", "model.layers.14.mlp.experts.58.down_proj.weight", "model.layers.14.mlp.experts.59.gate_and_up_proj.weight", "model.layers.14.mlp.experts.59.down_proj.weight", "model.layers.14.mlp.experts.60.gate_and_up_proj.weight", "model.layers.14.mlp.experts.60.down_proj.weight", "model.layers.14.mlp.experts.61.gate_and_up_proj.weight", "model.layers.14.mlp.experts.61.down_proj.weight", "model.layers.14.mlp.experts.62.gate_and_up_proj.weight", "model.layers.14.mlp.experts.62.down_proj.weight", "model.layers.14.mlp.experts.63.gate_and_up_proj.weight", "model.layers.14.mlp.experts.63.down_proj.weight", "model.layers.15.self_attn.qkv_proj.weight", "model.layers.15.self_attn.o_proj.weight", "model.layers.15.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.15.mlp.shared_mlp.down_proj.weight", "model.layers.15.mlp.experts.0.gate_and_up_proj.weight", "model.layers.15.mlp.experts.0.down_proj.weight", "model.layers.15.mlp.experts.1.gate_and_up_proj.weight", "model.layers.15.mlp.experts.1.down_proj.weight", "model.layers.15.mlp.experts.2.gate_and_up_proj.weight", "model.layers.15.mlp.experts.2.down_proj.weight", "model.layers.15.mlp.experts.3.gate_and_up_proj.weight", "model.layers.15.mlp.experts.3.down_proj.weight", "model.layers.15.mlp.experts.4.gate_and_up_proj.weight", "model.layers.15.mlp.experts.4.down_proj.weight", "model.layers.15.mlp.experts.5.gate_and_up_proj.weight", "model.layers.15.mlp.experts.5.down_proj.weight", "model.layers.15.mlp.experts.6.gate_and_up_proj.weight", "model.layers.15.mlp.experts.6.down_proj.weight", "model.layers.15.mlp.experts.7.gate_and_up_proj.weight", "model.layers.15.mlp.experts.7.down_proj.weight", "model.layers.15.mlp.experts.8.gate_and_up_proj.weight", "model.layers.15.mlp.experts.8.down_proj.weight", "model.layers.15.mlp.experts.9.gate_and_up_proj.weight", "model.layers.15.mlp.experts.9.down_proj.weight", "model.layers.15.mlp.experts.10.gate_and_up_proj.weight", "model.layers.15.mlp.experts.10.down_proj.weight", "model.layers.15.mlp.experts.11.gate_and_up_proj.weight", "model.layers.15.mlp.experts.11.down_proj.weight", "model.layers.15.mlp.experts.12.gate_and_up_proj.weight", "model.layers.15.mlp.experts.12.down_proj.weight", "model.layers.15.mlp.experts.13.gate_and_up_proj.weight", "model.layers.15.mlp.experts.13.down_proj.weight", "model.layers.15.mlp.experts.14.gate_and_up_proj.weight", "model.layers.15.mlp.experts.14.down_proj.weight", "model.layers.15.mlp.experts.15.gate_and_up_proj.weight", "model.layers.15.mlp.experts.15.down_proj.weight", "model.layers.15.mlp.experts.16.gate_and_up_proj.weight", "model.layers.15.mlp.experts.16.down_proj.weight", "model.layers.15.mlp.experts.17.gate_and_up_proj.weight", "model.layers.15.mlp.experts.17.down_proj.weight", "model.layers.15.mlp.experts.18.gate_and_up_proj.weight", "model.layers.15.mlp.experts.18.down_proj.weight", "model.layers.15.mlp.experts.19.gate_and_up_proj.weight", "model.layers.15.mlp.experts.19.down_proj.weight", "model.layers.15.mlp.experts.20.gate_and_up_proj.weight", "model.layers.15.mlp.experts.20.down_proj.weight", "model.layers.15.mlp.experts.21.gate_and_up_proj.weight", "model.layers.15.mlp.experts.21.down_proj.weight", "model.layers.15.mlp.experts.22.gate_and_up_proj.weight", "model.layers.15.mlp.experts.22.down_proj.weight", "model.layers.15.mlp.experts.23.gate_and_up_proj.weight", "model.layers.15.mlp.experts.23.down_proj.weight", "model.layers.15.mlp.experts.24.gate_and_up_proj.weight", "model.layers.15.mlp.experts.24.down_proj.weight", "model.layers.15.mlp.experts.25.gate_and_up_proj.weight", "model.layers.15.mlp.experts.25.down_proj.weight", "model.layers.15.mlp.experts.26.gate_and_up_proj.weight", "model.layers.15.mlp.experts.26.down_proj.weight", "model.layers.15.mlp.experts.27.gate_and_up_proj.weight", "model.layers.15.mlp.experts.27.down_proj.weight", "model.layers.15.mlp.experts.28.gate_and_up_proj.weight", "model.layers.15.mlp.experts.28.down_proj.weight", "model.layers.15.mlp.experts.29.gate_and_up_proj.weight", "model.layers.15.mlp.experts.29.down_proj.weight", "model.layers.15.mlp.experts.30.gate_and_up_proj.weight", "model.layers.15.mlp.experts.30.down_proj.weight", "model.layers.15.mlp.experts.31.gate_and_up_proj.weight", "model.layers.15.mlp.experts.31.down_proj.weight", "model.layers.15.mlp.experts.32.gate_and_up_proj.weight", "model.layers.15.mlp.experts.32.down_proj.weight", "model.layers.15.mlp.experts.33.gate_and_up_proj.weight", "model.layers.15.mlp.experts.33.down_proj.weight", "model.layers.15.mlp.experts.34.gate_and_up_proj.weight", "model.layers.15.mlp.experts.34.down_proj.weight", "model.layers.15.mlp.experts.35.gate_and_up_proj.weight", "model.layers.15.mlp.experts.35.down_proj.weight", "model.layers.15.mlp.experts.36.gate_and_up_proj.weight", "model.layers.15.mlp.experts.36.down_proj.weight", "model.layers.15.mlp.experts.37.gate_and_up_proj.weight", "model.layers.15.mlp.experts.37.down_proj.weight", "model.layers.15.mlp.experts.38.gate_and_up_proj.weight", "model.layers.15.mlp.experts.38.down_proj.weight", "model.layers.15.mlp.experts.39.gate_and_up_proj.weight", "model.layers.15.mlp.experts.39.down_proj.weight", "model.layers.15.mlp.experts.40.gate_and_up_proj.weight", "model.layers.15.mlp.experts.40.down_proj.weight", "model.layers.15.mlp.experts.41.gate_and_up_proj.weight", "model.layers.15.mlp.experts.41.down_proj.weight", "model.layers.15.mlp.experts.42.gate_and_up_proj.weight", "model.layers.15.mlp.experts.42.down_proj.weight", "model.layers.15.mlp.experts.43.gate_and_up_proj.weight", "model.layers.15.mlp.experts.43.down_proj.weight", "model.layers.15.mlp.experts.44.gate_and_up_proj.weight", "model.layers.15.mlp.experts.44.down_proj.weight", "model.layers.15.mlp.experts.45.gate_and_up_proj.weight", "model.layers.15.mlp.experts.45.down_proj.weight", "model.layers.15.mlp.experts.46.gate_and_up_proj.weight", "model.layers.15.mlp.experts.46.down_proj.weight", "model.layers.15.mlp.experts.47.gate_and_up_proj.weight", "model.layers.15.mlp.experts.47.down_proj.weight", "model.layers.15.mlp.experts.48.gate_and_up_proj.weight", "model.layers.15.mlp.experts.48.down_proj.weight", "model.layers.15.mlp.experts.49.gate_and_up_proj.weight", "model.layers.15.mlp.experts.49.down_proj.weight", "model.layers.15.mlp.experts.50.gate_and_up_proj.weight", "model.layers.15.mlp.experts.50.down_proj.weight", "model.layers.15.mlp.experts.51.gate_and_up_proj.weight", "model.layers.15.mlp.experts.51.down_proj.weight", "model.layers.15.mlp.experts.52.gate_and_up_proj.weight", "model.layers.15.mlp.experts.52.down_proj.weight", "model.layers.15.mlp.experts.53.gate_and_up_proj.weight", "model.layers.15.mlp.experts.53.down_proj.weight", "model.layers.15.mlp.experts.54.gate_and_up_proj.weight", "model.layers.15.mlp.experts.54.down_proj.weight", "model.layers.15.mlp.experts.55.gate_and_up_proj.weight", "model.layers.15.mlp.experts.55.down_proj.weight", "model.layers.15.mlp.experts.56.gate_and_up_proj.weight", "model.layers.15.mlp.experts.56.down_proj.weight", "model.layers.15.mlp.experts.57.gate_and_up_proj.weight", "model.layers.15.mlp.experts.57.down_proj.weight", "model.layers.15.mlp.experts.58.gate_and_up_proj.weight", "model.layers.15.mlp.experts.58.down_proj.weight", "model.layers.15.mlp.experts.59.gate_and_up_proj.weight", "model.layers.15.mlp.experts.59.down_proj.weight", "model.layers.15.mlp.experts.60.gate_and_up_proj.weight", "model.layers.15.mlp.experts.60.down_proj.weight", "model.layers.15.mlp.experts.61.gate_and_up_proj.weight", "model.layers.15.mlp.experts.61.down_proj.weight", "model.layers.15.mlp.experts.62.gate_and_up_proj.weight", "model.layers.15.mlp.experts.62.down_proj.weight", "model.layers.15.mlp.experts.63.gate_and_up_proj.weight", "model.layers.15.mlp.experts.63.down_proj.weight", "model.layers.16.self_attn.qkv_proj.weight", "model.layers.16.self_attn.o_proj.weight", "model.layers.16.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.16.mlp.shared_mlp.down_proj.weight", "model.layers.16.mlp.experts.0.gate_and_up_proj.weight", "model.layers.16.mlp.experts.0.down_proj.weight", "model.layers.16.mlp.experts.1.gate_and_up_proj.weight", "model.layers.16.mlp.experts.1.down_proj.weight", "model.layers.16.mlp.experts.2.gate_and_up_proj.weight", "model.layers.16.mlp.experts.2.down_proj.weight", "model.layers.16.mlp.experts.3.gate_and_up_proj.weight", "model.layers.16.mlp.experts.3.down_proj.weight", "model.layers.16.mlp.experts.4.gate_and_up_proj.weight", "model.layers.16.mlp.experts.4.down_proj.weight", "model.layers.16.mlp.experts.5.gate_and_up_proj.weight", "model.layers.16.mlp.experts.5.down_proj.weight", "model.layers.16.mlp.experts.6.gate_and_up_proj.weight", "model.layers.16.mlp.experts.6.down_proj.weight", "model.layers.16.mlp.experts.7.gate_and_up_proj.weight", "model.layers.16.mlp.experts.7.down_proj.weight", "model.layers.16.mlp.experts.8.gate_and_up_proj.weight", "model.layers.16.mlp.experts.8.down_proj.weight", "model.layers.16.mlp.experts.9.gate_and_up_proj.weight", "model.layers.16.mlp.experts.9.down_proj.weight", "model.layers.16.mlp.experts.10.gate_and_up_proj.weight", "model.layers.16.mlp.experts.10.down_proj.weight", "model.layers.16.mlp.experts.11.gate_and_up_proj.weight", "model.layers.16.mlp.experts.11.down_proj.weight", "model.layers.16.mlp.experts.12.gate_and_up_proj.weight", "model.layers.16.mlp.experts.12.down_proj.weight", "model.layers.16.mlp.experts.13.gate_and_up_proj.weight", "model.layers.16.mlp.experts.13.down_proj.weight", "model.layers.16.mlp.experts.14.gate_and_up_proj.weight", "model.layers.16.mlp.experts.14.down_proj.weight", "model.layers.16.mlp.experts.15.gate_and_up_proj.weight", "model.layers.16.mlp.experts.15.down_proj.weight", "model.layers.16.mlp.experts.16.gate_and_up_proj.weight", "model.layers.16.mlp.experts.16.down_proj.weight", "model.layers.16.mlp.experts.17.gate_and_up_proj.weight", "model.layers.16.mlp.experts.17.down_proj.weight", "model.layers.16.mlp.experts.18.gate_and_up_proj.weight", "model.layers.16.mlp.experts.18.down_proj.weight", "model.layers.16.mlp.experts.19.gate_and_up_proj.weight", "model.layers.16.mlp.experts.19.down_proj.weight", "model.layers.16.mlp.experts.20.gate_and_up_proj.weight", "model.layers.16.mlp.experts.20.down_proj.weight", "model.layers.16.mlp.experts.21.gate_and_up_proj.weight", "model.layers.16.mlp.experts.21.down_proj.weight", "model.layers.16.mlp.experts.22.gate_and_up_proj.weight", "model.layers.16.mlp.experts.22.down_proj.weight", "model.layers.16.mlp.experts.23.gate_and_up_proj.weight", "model.layers.16.mlp.experts.23.down_proj.weight", "model.layers.16.mlp.experts.24.gate_and_up_proj.weight", "model.layers.16.mlp.experts.24.down_proj.weight", "model.layers.16.mlp.experts.25.gate_and_up_proj.weight", "model.layers.16.mlp.experts.25.down_proj.weight", "model.layers.16.mlp.experts.26.gate_and_up_proj.weight", "model.layers.16.mlp.experts.26.down_proj.weight", "model.layers.16.mlp.experts.27.gate_and_up_proj.weight", "model.layers.16.mlp.experts.27.down_proj.weight", "model.layers.16.mlp.experts.28.gate_and_up_proj.weight", "model.layers.16.mlp.experts.28.down_proj.weight", "model.layers.16.mlp.experts.29.gate_and_up_proj.weight", "model.layers.16.mlp.experts.29.down_proj.weight", "model.layers.16.mlp.experts.30.gate_and_up_proj.weight", "model.layers.16.mlp.experts.30.down_proj.weight", "model.layers.16.mlp.experts.31.gate_and_up_proj.weight", "model.layers.16.mlp.experts.31.down_proj.weight", "model.layers.16.mlp.experts.32.gate_and_up_proj.weight", "model.layers.16.mlp.experts.32.down_proj.weight", "model.layers.16.mlp.experts.33.gate_and_up_proj.weight", "model.layers.16.mlp.experts.33.down_proj.weight", "model.layers.16.mlp.experts.34.gate_and_up_proj.weight", "model.layers.16.mlp.experts.34.down_proj.weight", "model.layers.16.mlp.experts.35.gate_and_up_proj.weight", "model.layers.16.mlp.experts.35.down_proj.weight", "model.layers.16.mlp.experts.36.gate_and_up_proj.weight", "model.layers.16.mlp.experts.36.down_proj.weight", "model.layers.16.mlp.experts.37.gate_and_up_proj.weight", "model.layers.16.mlp.experts.37.down_proj.weight", "model.layers.16.mlp.experts.38.gate_and_up_proj.weight", "model.layers.16.mlp.experts.38.down_proj.weight", "model.layers.16.mlp.experts.39.gate_and_up_proj.weight", "model.layers.16.mlp.experts.39.down_proj.weight", "model.layers.16.mlp.experts.40.gate_and_up_proj.weight", "model.layers.16.mlp.experts.40.down_proj.weight", "model.layers.16.mlp.experts.41.gate_and_up_proj.weight", "model.layers.16.mlp.experts.41.down_proj.weight", "model.layers.16.mlp.experts.42.gate_and_up_proj.weight", "model.layers.16.mlp.experts.42.down_proj.weight", "model.layers.16.mlp.experts.43.gate_and_up_proj.weight", "model.layers.16.mlp.experts.43.down_proj.weight", "model.layers.16.mlp.experts.44.gate_and_up_proj.weight", "model.layers.16.mlp.experts.44.down_proj.weight", "model.layers.16.mlp.experts.45.gate_and_up_proj.weight", "model.layers.16.mlp.experts.45.down_proj.weight", "model.layers.16.mlp.experts.46.gate_and_up_proj.weight", "model.layers.16.mlp.experts.46.down_proj.weight", "model.layers.16.mlp.experts.47.gate_and_up_proj.weight", "model.layers.16.mlp.experts.47.down_proj.weight", "model.layers.16.mlp.experts.48.gate_and_up_proj.weight", "model.layers.16.mlp.experts.48.down_proj.weight", "model.layers.16.mlp.experts.49.gate_and_up_proj.weight", "model.layers.16.mlp.experts.49.down_proj.weight", "model.layers.16.mlp.experts.50.gate_and_up_proj.weight", "model.layers.16.mlp.experts.50.down_proj.weight", "model.layers.16.mlp.experts.51.gate_and_up_proj.weight", "model.layers.16.mlp.experts.51.down_proj.weight", "model.layers.16.mlp.experts.52.gate_and_up_proj.weight", "model.layers.16.mlp.experts.52.down_proj.weight", "model.layers.16.mlp.experts.53.gate_and_up_proj.weight", "model.layers.16.mlp.experts.53.down_proj.weight", "model.layers.16.mlp.experts.54.gate_and_up_proj.weight", "model.layers.16.mlp.experts.54.down_proj.weight", "model.layers.16.mlp.experts.55.gate_and_up_proj.weight", "model.layers.16.mlp.experts.55.down_proj.weight", "model.layers.16.mlp.experts.56.gate_and_up_proj.weight", "model.layers.16.mlp.experts.56.down_proj.weight", "model.layers.16.mlp.experts.57.gate_and_up_proj.weight", "model.layers.16.mlp.experts.57.down_proj.weight", "model.layers.16.mlp.experts.58.gate_and_up_proj.weight", "model.layers.16.mlp.experts.58.down_proj.weight", "model.layers.16.mlp.experts.59.gate_and_up_proj.weight", "model.layers.16.mlp.experts.59.down_proj.weight", "model.layers.16.mlp.experts.60.gate_and_up_proj.weight", "model.layers.16.mlp.experts.60.down_proj.weight", "model.layers.16.mlp.experts.61.gate_and_up_proj.weight", "model.layers.16.mlp.experts.61.down_proj.weight", "model.layers.16.mlp.experts.62.gate_and_up_proj.weight", "model.layers.16.mlp.experts.62.down_proj.weight", "model.layers.16.mlp.experts.63.gate_and_up_proj.weight", "model.layers.16.mlp.experts.63.down_proj.weight", "model.layers.17.self_attn.qkv_proj.weight", "model.layers.17.self_attn.o_proj.weight", "model.layers.17.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.17.mlp.shared_mlp.down_proj.weight", "model.layers.17.mlp.experts.0.gate_and_up_proj.weight", "model.layers.17.mlp.experts.0.down_proj.weight", "model.layers.17.mlp.experts.1.gate_and_up_proj.weight", "model.layers.17.mlp.experts.1.down_proj.weight", "model.layers.17.mlp.experts.2.gate_and_up_proj.weight", "model.layers.17.mlp.experts.2.down_proj.weight", "model.layers.17.mlp.experts.3.gate_and_up_proj.weight", "model.layers.17.mlp.experts.3.down_proj.weight", "model.layers.17.mlp.experts.4.gate_and_up_proj.weight", "model.layers.17.mlp.experts.4.down_proj.weight", "model.layers.17.mlp.experts.5.gate_and_up_proj.weight", "model.layers.17.mlp.experts.5.down_proj.weight", "model.layers.17.mlp.experts.6.gate_and_up_proj.weight", "model.layers.17.mlp.experts.6.down_proj.weight", "model.layers.17.mlp.experts.7.gate_and_up_proj.weight", "model.layers.17.mlp.experts.7.down_proj.weight", "model.layers.17.mlp.experts.8.gate_and_up_proj.weight", "model.layers.17.mlp.experts.8.down_proj.weight", "model.layers.17.mlp.experts.9.gate_and_up_proj.weight", "model.layers.17.mlp.experts.9.down_proj.weight", "model.layers.17.mlp.experts.10.gate_and_up_proj.weight", "model.layers.17.mlp.experts.10.down_proj.weight", "model.layers.17.mlp.experts.11.gate_and_up_proj.weight", "model.layers.17.mlp.experts.11.down_proj.weight", "model.layers.17.mlp.experts.12.gate_and_up_proj.weight", "model.layers.17.mlp.experts.12.down_proj.weight", "model.layers.17.mlp.experts.13.gate_and_up_proj.weight", "model.layers.17.mlp.experts.13.down_proj.weight", "model.layers.17.mlp.experts.14.gate_and_up_proj.weight", "model.layers.17.mlp.experts.14.down_proj.weight", "model.layers.17.mlp.experts.15.gate_and_up_proj.weight", "model.layers.17.mlp.experts.15.down_proj.weight", "model.layers.17.mlp.experts.16.gate_and_up_proj.weight", "model.layers.17.mlp.experts.16.down_proj.weight", "model.layers.17.mlp.experts.17.gate_and_up_proj.weight", "model.layers.17.mlp.experts.17.down_proj.weight", "model.layers.17.mlp.experts.18.gate_and_up_proj.weight", "model.layers.17.mlp.experts.18.down_proj.weight", "model.layers.17.mlp.experts.19.gate_and_up_proj.weight", "model.layers.17.mlp.experts.19.down_proj.weight", "model.layers.17.mlp.experts.20.gate_and_up_proj.weight", "model.layers.17.mlp.experts.20.down_proj.weight", "model.layers.17.mlp.experts.21.gate_and_up_proj.weight", "model.layers.17.mlp.experts.21.down_proj.weight", "model.layers.17.mlp.experts.22.gate_and_up_proj.weight", "model.layers.17.mlp.experts.22.down_proj.weight", "model.layers.17.mlp.experts.23.gate_and_up_proj.weight", "model.layers.17.mlp.experts.23.down_proj.weight", "model.layers.17.mlp.experts.24.gate_and_up_proj.weight", "model.layers.17.mlp.experts.24.down_proj.weight", "model.layers.17.mlp.experts.25.gate_and_up_proj.weight", "model.layers.17.mlp.experts.25.down_proj.weight", "model.layers.17.mlp.experts.26.gate_and_up_proj.weight", "model.layers.17.mlp.experts.26.down_proj.weight", "model.layers.17.mlp.experts.27.gate_and_up_proj.weight", "model.layers.17.mlp.experts.27.down_proj.weight", "model.layers.17.mlp.experts.28.gate_and_up_proj.weight", "model.layers.17.mlp.experts.28.down_proj.weight", "model.layers.17.mlp.experts.29.gate_and_up_proj.weight", "model.layers.17.mlp.experts.29.down_proj.weight", "model.layers.17.mlp.experts.30.gate_and_up_proj.weight", "model.layers.17.mlp.experts.30.down_proj.weight", "model.layers.17.mlp.experts.31.gate_and_up_proj.weight", "model.layers.17.mlp.experts.31.down_proj.weight", "model.layers.17.mlp.experts.32.gate_and_up_proj.weight", "model.layers.17.mlp.experts.32.down_proj.weight", "model.layers.17.mlp.experts.33.gate_and_up_proj.weight", "model.layers.17.mlp.experts.33.down_proj.weight", "model.layers.17.mlp.experts.34.gate_and_up_proj.weight", "model.layers.17.mlp.experts.34.down_proj.weight", "model.layers.17.mlp.experts.35.gate_and_up_proj.weight", "model.layers.17.mlp.experts.35.down_proj.weight", "model.layers.17.mlp.experts.36.gate_and_up_proj.weight", "model.layers.17.mlp.experts.36.down_proj.weight", "model.layers.17.mlp.experts.37.gate_and_up_proj.weight", "model.layers.17.mlp.experts.37.down_proj.weight", "model.layers.17.mlp.experts.38.gate_and_up_proj.weight", "model.layers.17.mlp.experts.38.down_proj.weight", "model.layers.17.mlp.experts.39.gate_and_up_proj.weight", "model.layers.17.mlp.experts.39.down_proj.weight", "model.layers.17.mlp.experts.40.gate_and_up_proj.weight", "model.layers.17.mlp.experts.40.down_proj.weight", "model.layers.17.mlp.experts.41.gate_and_up_proj.weight", "model.layers.17.mlp.experts.41.down_proj.weight", "model.layers.17.mlp.experts.42.gate_and_up_proj.weight", "model.layers.17.mlp.experts.42.down_proj.weight", "model.layers.17.mlp.experts.43.gate_and_up_proj.weight", "model.layers.17.mlp.experts.43.down_proj.weight", "model.layers.17.mlp.experts.44.gate_and_up_proj.weight", "model.layers.17.mlp.experts.44.down_proj.weight", "model.layers.17.mlp.experts.45.gate_and_up_proj.weight", "model.layers.17.mlp.experts.45.down_proj.weight", "model.layers.17.mlp.experts.46.gate_and_up_proj.weight", "model.layers.17.mlp.experts.46.down_proj.weight", "model.layers.17.mlp.experts.47.gate_and_up_proj.weight", "model.layers.17.mlp.experts.47.down_proj.weight", "model.layers.17.mlp.experts.48.gate_and_up_proj.weight", "model.layers.17.mlp.experts.48.down_proj.weight", "model.layers.17.mlp.experts.49.gate_and_up_proj.weight", "model.layers.17.mlp.experts.49.down_proj.weight", "model.layers.17.mlp.experts.50.gate_and_up_proj.weight", "model.layers.17.mlp.experts.50.down_proj.weight", "model.layers.17.mlp.experts.51.gate_and_up_proj.weight", "model.layers.17.mlp.experts.51.down_proj.weight", "model.layers.17.mlp.experts.52.gate_and_up_proj.weight", "model.layers.17.mlp.experts.52.down_proj.weight", "model.layers.17.mlp.experts.53.gate_and_up_proj.weight", "model.layers.17.mlp.experts.53.down_proj.weight", "model.layers.17.mlp.experts.54.gate_and_up_proj.weight", "model.layers.17.mlp.experts.54.down_proj.weight", "model.layers.17.mlp.experts.55.gate_and_up_proj.weight", "model.layers.17.mlp.experts.55.down_proj.weight", "model.layers.17.mlp.experts.56.gate_and_up_proj.weight", "model.layers.17.mlp.experts.56.down_proj.weight", "model.layers.17.mlp.experts.57.gate_and_up_proj.weight", "model.layers.17.mlp.experts.57.down_proj.weight", "model.layers.17.mlp.experts.58.gate_and_up_proj.weight", "model.layers.17.mlp.experts.58.down_proj.weight", "model.layers.17.mlp.experts.59.gate_and_up_proj.weight", "model.layers.17.mlp.experts.59.down_proj.weight", "model.layers.17.mlp.experts.60.gate_and_up_proj.weight", "model.layers.17.mlp.experts.60.down_proj.weight", "model.layers.17.mlp.experts.61.gate_and_up_proj.weight", "model.layers.17.mlp.experts.61.down_proj.weight", "model.layers.17.mlp.experts.62.gate_and_up_proj.weight", "model.layers.17.mlp.experts.62.down_proj.weight", "model.layers.17.mlp.experts.63.gate_and_up_proj.weight", "model.layers.17.mlp.experts.63.down_proj.weight", "model.layers.18.self_attn.qkv_proj.weight", "model.layers.18.self_attn.o_proj.weight", "model.layers.18.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.18.mlp.shared_mlp.down_proj.weight", "model.layers.18.mlp.experts.0.gate_and_up_proj.weight", "model.layers.18.mlp.experts.0.down_proj.weight", "model.layers.18.mlp.experts.1.gate_and_up_proj.weight", "model.layers.18.mlp.experts.1.down_proj.weight", "model.layers.18.mlp.experts.2.gate_and_up_proj.weight", "model.layers.18.mlp.experts.2.down_proj.weight", "model.layers.18.mlp.experts.3.gate_and_up_proj.weight", "model.layers.18.mlp.experts.3.down_proj.weight", "model.layers.18.mlp.experts.4.gate_and_up_proj.weight", "model.layers.18.mlp.experts.4.down_proj.weight", "model.layers.18.mlp.experts.5.gate_and_up_proj.weight", "model.layers.18.mlp.experts.5.down_proj.weight", "model.layers.18.mlp.experts.6.gate_and_up_proj.weight", "model.layers.18.mlp.experts.6.down_proj.weight", "model.layers.18.mlp.experts.7.gate_and_up_proj.weight", "model.layers.18.mlp.experts.7.down_proj.weight", "model.layers.18.mlp.experts.8.gate_and_up_proj.weight", "model.layers.18.mlp.experts.8.down_proj.weight", "model.layers.18.mlp.experts.9.gate_and_up_proj.weight", "model.layers.18.mlp.experts.9.down_proj.weight", "model.layers.18.mlp.experts.10.gate_and_up_proj.weight", "model.layers.18.mlp.experts.10.down_proj.weight", "model.layers.18.mlp.experts.11.gate_and_up_proj.weight", "model.layers.18.mlp.experts.11.down_proj.weight", "model.layers.18.mlp.experts.12.gate_and_up_proj.weight", "model.layers.18.mlp.experts.12.down_proj.weight", "model.layers.18.mlp.experts.13.gate_and_up_proj.weight", "model.layers.18.mlp.experts.13.down_proj.weight", "model.layers.18.mlp.experts.14.gate_and_up_proj.weight", "model.layers.18.mlp.experts.14.down_proj.weight", "model.layers.18.mlp.experts.15.gate_and_up_proj.weight", "model.layers.18.mlp.experts.15.down_proj.weight", "model.layers.18.mlp.experts.16.gate_and_up_proj.weight", "model.layers.18.mlp.experts.16.down_proj.weight", "model.layers.18.mlp.experts.17.gate_and_up_proj.weight", "model.layers.18.mlp.experts.17.down_proj.weight", "model.layers.18.mlp.experts.18.gate_and_up_proj.weight", "model.layers.18.mlp.experts.18.down_proj.weight", "model.layers.18.mlp.experts.19.gate_and_up_proj.weight", "model.layers.18.mlp.experts.19.down_proj.weight", "model.layers.18.mlp.experts.20.gate_and_up_proj.weight", "model.layers.18.mlp.experts.20.down_proj.weight", "model.layers.18.mlp.experts.21.gate_and_up_proj.weight", "model.layers.18.mlp.experts.21.down_proj.weight", "model.layers.18.mlp.experts.22.gate_and_up_proj.weight", "model.layers.18.mlp.experts.22.down_proj.weight", "model.layers.18.mlp.experts.23.gate_and_up_proj.weight", "model.layers.18.mlp.experts.23.down_proj.weight", "model.layers.18.mlp.experts.24.gate_and_up_proj.weight", "model.layers.18.mlp.experts.24.down_proj.weight", "model.layers.18.mlp.experts.25.gate_and_up_proj.weight", "model.layers.18.mlp.experts.25.down_proj.weight", "model.layers.18.mlp.experts.26.gate_and_up_proj.weight", "model.layers.18.mlp.experts.26.down_proj.weight", "model.layers.18.mlp.experts.27.gate_and_up_proj.weight", "model.layers.18.mlp.experts.27.down_proj.weight", "model.layers.18.mlp.experts.28.gate_and_up_proj.weight", "model.layers.18.mlp.experts.28.down_proj.weight", "model.layers.18.mlp.experts.29.gate_and_up_proj.weight", "model.layers.18.mlp.experts.29.down_proj.weight", "model.layers.18.mlp.experts.30.gate_and_up_proj.weight", "model.layers.18.mlp.experts.30.down_proj.weight", "model.layers.18.mlp.experts.31.gate_and_up_proj.weight", "model.layers.18.mlp.experts.31.down_proj.weight", "model.layers.18.mlp.experts.32.gate_and_up_proj.weight", "model.layers.18.mlp.experts.32.down_proj.weight", "model.layers.18.mlp.experts.33.gate_and_up_proj.weight", "model.layers.18.mlp.experts.33.down_proj.weight", "model.layers.18.mlp.experts.34.gate_and_up_proj.weight", "model.layers.18.mlp.experts.34.down_proj.weight", "model.layers.18.mlp.experts.35.gate_and_up_proj.weight", "model.layers.18.mlp.experts.35.down_proj.weight", "model.layers.18.mlp.experts.36.gate_and_up_proj.weight", "model.layers.18.mlp.experts.36.down_proj.weight", "model.layers.18.mlp.experts.37.gate_and_up_proj.weight", "model.layers.18.mlp.experts.37.down_proj.weight", "model.layers.18.mlp.experts.38.gate_and_up_proj.weight", "model.layers.18.mlp.experts.38.down_proj.weight", "model.layers.18.mlp.experts.39.gate_and_up_proj.weight", "model.layers.18.mlp.experts.39.down_proj.weight", "model.layers.18.mlp.experts.40.gate_and_up_proj.weight", "model.layers.18.mlp.experts.40.down_proj.weight", "model.layers.18.mlp.experts.41.gate_and_up_proj.weight", "model.layers.18.mlp.experts.41.down_proj.weight", "model.layers.18.mlp.experts.42.gate_and_up_proj.weight", "model.layers.18.mlp.experts.42.down_proj.weight", "model.layers.18.mlp.experts.43.gate_and_up_proj.weight", "model.layers.18.mlp.experts.43.down_proj.weight", "model.layers.18.mlp.experts.44.gate_and_up_proj.weight", "model.layers.18.mlp.experts.44.down_proj.weight", "model.layers.18.mlp.experts.45.gate_and_up_proj.weight", "model.layers.18.mlp.experts.45.down_proj.weight", "model.layers.18.mlp.experts.46.gate_and_up_proj.weight", "model.layers.18.mlp.experts.46.down_proj.weight", "model.layers.18.mlp.experts.47.gate_and_up_proj.weight", "model.layers.18.mlp.experts.47.down_proj.weight", "model.layers.18.mlp.experts.48.gate_and_up_proj.weight", "model.layers.18.mlp.experts.48.down_proj.weight", "model.layers.18.mlp.experts.49.gate_and_up_proj.weight", "model.layers.18.mlp.experts.49.down_proj.weight", "model.layers.18.mlp.experts.50.gate_and_up_proj.weight", "model.layers.18.mlp.experts.50.down_proj.weight", "model.layers.18.mlp.experts.51.gate_and_up_proj.weight", "model.layers.18.mlp.experts.51.down_proj.weight", "model.layers.18.mlp.experts.52.gate_and_up_proj.weight", "model.layers.18.mlp.experts.52.down_proj.weight", "model.layers.18.mlp.experts.53.gate_and_up_proj.weight", "model.layers.18.mlp.experts.53.down_proj.weight", "model.layers.18.mlp.experts.54.gate_and_up_proj.weight", "model.layers.18.mlp.experts.54.down_proj.weight", "model.layers.18.mlp.experts.55.gate_and_up_proj.weight", "model.layers.18.mlp.experts.55.down_proj.weight", "model.layers.18.mlp.experts.56.gate_and_up_proj.weight", "model.layers.18.mlp.experts.56.down_proj.weight", "model.layers.18.mlp.experts.57.gate_and_up_proj.weight", "model.layers.18.mlp.experts.57.down_proj.weight", "model.layers.18.mlp.experts.58.gate_and_up_proj.weight", "model.layers.18.mlp.experts.58.down_proj.weight", "model.layers.18.mlp.experts.59.gate_and_up_proj.weight", "model.layers.18.mlp.experts.59.down_proj.weight", "model.layers.18.mlp.experts.60.gate_and_up_proj.weight", "model.layers.18.mlp.experts.60.down_proj.weight", "model.layers.18.mlp.experts.61.gate_and_up_proj.weight", "model.layers.18.mlp.experts.61.down_proj.weight", "model.layers.18.mlp.experts.62.gate_and_up_proj.weight", "model.layers.18.mlp.experts.62.down_proj.weight", "model.layers.18.mlp.experts.63.gate_and_up_proj.weight", "model.layers.18.mlp.experts.63.down_proj.weight", "model.layers.19.self_attn.qkv_proj.weight", "model.layers.19.self_attn.o_proj.weight", "model.layers.19.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.19.mlp.shared_mlp.down_proj.weight", "model.layers.19.mlp.experts.0.gate_and_up_proj.weight", "model.layers.19.mlp.experts.0.down_proj.weight", "model.layers.19.mlp.experts.1.gate_and_up_proj.weight", "model.layers.19.mlp.experts.1.down_proj.weight", "model.layers.19.mlp.experts.2.gate_and_up_proj.weight", "model.layers.19.mlp.experts.2.down_proj.weight", "model.layers.19.mlp.experts.3.gate_and_up_proj.weight", "model.layers.19.mlp.experts.3.down_proj.weight", "model.layers.19.mlp.experts.4.gate_and_up_proj.weight", "model.layers.19.mlp.experts.4.down_proj.weight", "model.layers.19.mlp.experts.5.gate_and_up_proj.weight", "model.layers.19.mlp.experts.5.down_proj.weight", "model.layers.19.mlp.experts.6.gate_and_up_proj.weight", "model.layers.19.mlp.experts.6.down_proj.weight", "model.layers.19.mlp.experts.7.gate_and_up_proj.weight", "model.layers.19.mlp.experts.7.down_proj.weight", "model.layers.19.mlp.experts.8.gate_and_up_proj.weight", "model.layers.19.mlp.experts.8.down_proj.weight", "model.layers.19.mlp.experts.9.gate_and_up_proj.weight", "model.layers.19.mlp.experts.9.down_proj.weight", "model.layers.19.mlp.experts.10.gate_and_up_proj.weight", "model.layers.19.mlp.experts.10.down_proj.weight", "model.layers.19.mlp.experts.11.gate_and_up_proj.weight", "model.layers.19.mlp.experts.11.down_proj.weight", "model.layers.19.mlp.experts.12.gate_and_up_proj.weight", "model.layers.19.mlp.experts.12.down_proj.weight", "model.layers.19.mlp.experts.13.gate_and_up_proj.weight", "model.layers.19.mlp.experts.13.down_proj.weight", "model.layers.19.mlp.experts.14.gate_and_up_proj.weight", "model.layers.19.mlp.experts.14.down_proj.weight", "model.layers.19.mlp.experts.15.gate_and_up_proj.weight", "model.layers.19.mlp.experts.15.down_proj.weight", "model.layers.19.mlp.experts.16.gate_and_up_proj.weight", "model.layers.19.mlp.experts.16.down_proj.weight", "model.layers.19.mlp.experts.17.gate_and_up_proj.weight", "model.layers.19.mlp.experts.17.down_proj.weight", "model.layers.19.mlp.experts.18.gate_and_up_proj.weight", "model.layers.19.mlp.experts.18.down_proj.weight", "model.layers.19.mlp.experts.19.gate_and_up_proj.weight", "model.layers.19.mlp.experts.19.down_proj.weight", "model.layers.19.mlp.experts.20.gate_and_up_proj.weight", "model.layers.19.mlp.experts.20.down_proj.weight", "model.layers.19.mlp.experts.21.gate_and_up_proj.weight", "model.layers.19.mlp.experts.21.down_proj.weight", "model.layers.19.mlp.experts.22.gate_and_up_proj.weight", "model.layers.19.mlp.experts.22.down_proj.weight", "model.layers.19.mlp.experts.23.gate_and_up_proj.weight", "model.layers.19.mlp.experts.23.down_proj.weight", "model.layers.19.mlp.experts.24.gate_and_up_proj.weight", "model.layers.19.mlp.experts.24.down_proj.weight", "model.layers.19.mlp.experts.25.gate_and_up_proj.weight", "model.layers.19.mlp.experts.25.down_proj.weight", "model.layers.19.mlp.experts.26.gate_and_up_proj.weight", "model.layers.19.mlp.experts.26.down_proj.weight", "model.layers.19.mlp.experts.27.gate_and_up_proj.weight", "model.layers.19.mlp.experts.27.down_proj.weight", "model.layers.19.mlp.experts.28.gate_and_up_proj.weight", "model.layers.19.mlp.experts.28.down_proj.weight", "model.layers.19.mlp.experts.29.gate_and_up_proj.weight", "model.layers.19.mlp.experts.29.down_proj.weight", "model.layers.19.mlp.experts.30.gate_and_up_proj.weight", "model.layers.19.mlp.experts.30.down_proj.weight", "model.layers.19.mlp.experts.31.gate_and_up_proj.weight", "model.layers.19.mlp.experts.31.down_proj.weight", "model.layers.19.mlp.experts.32.gate_and_up_proj.weight", "model.layers.19.mlp.experts.32.down_proj.weight", "model.layers.19.mlp.experts.33.gate_and_up_proj.weight", "model.layers.19.mlp.experts.33.down_proj.weight", "model.layers.19.mlp.experts.34.gate_and_up_proj.weight", "model.layers.19.mlp.experts.34.down_proj.weight", "model.layers.19.mlp.experts.35.gate_and_up_proj.weight", "model.layers.19.mlp.experts.35.down_proj.weight", "model.layers.19.mlp.experts.36.gate_and_up_proj.weight", "model.layers.19.mlp.experts.36.down_proj.weight", "model.layers.19.mlp.experts.37.gate_and_up_proj.weight", "model.layers.19.mlp.experts.37.down_proj.weight", "model.layers.19.mlp.experts.38.gate_and_up_proj.weight", "model.layers.19.mlp.experts.38.down_proj.weight", "model.layers.19.mlp.experts.39.gate_and_up_proj.weight", "model.layers.19.mlp.experts.39.down_proj.weight", "model.layers.19.mlp.experts.40.gate_and_up_proj.weight", "model.layers.19.mlp.experts.40.down_proj.weight", "model.layers.19.mlp.experts.41.gate_and_up_proj.weight", "model.layers.19.mlp.experts.41.down_proj.weight", "model.layers.19.mlp.experts.42.gate_and_up_proj.weight", "model.layers.19.mlp.experts.42.down_proj.weight", "model.layers.19.mlp.experts.43.gate_and_up_proj.weight", "model.layers.19.mlp.experts.43.down_proj.weight", "model.layers.19.mlp.experts.44.gate_and_up_proj.weight", "model.layers.19.mlp.experts.44.down_proj.weight", "model.layers.19.mlp.experts.45.gate_and_up_proj.weight", "model.layers.19.mlp.experts.45.down_proj.weight", "model.layers.19.mlp.experts.46.gate_and_up_proj.weight", "model.layers.19.mlp.experts.46.down_proj.weight", "model.layers.19.mlp.experts.47.gate_and_up_proj.weight", "model.layers.19.mlp.experts.47.down_proj.weight", "model.layers.19.mlp.experts.48.gate_and_up_proj.weight", "model.layers.19.mlp.experts.48.down_proj.weight", "model.layers.19.mlp.experts.49.gate_and_up_proj.weight", "model.layers.19.mlp.experts.49.down_proj.weight", "model.layers.19.mlp.experts.50.gate_and_up_proj.weight", "model.layers.19.mlp.experts.50.down_proj.weight", "model.layers.19.mlp.experts.51.gate_and_up_proj.weight", "model.layers.19.mlp.experts.51.down_proj.weight", "model.layers.19.mlp.experts.52.gate_and_up_proj.weight", "model.layers.19.mlp.experts.52.down_proj.weight", "model.layers.19.mlp.experts.53.gate_and_up_proj.weight", "model.layers.19.mlp.experts.53.down_proj.weight", "model.layers.19.mlp.experts.54.gate_and_up_proj.weight", "model.layers.19.mlp.experts.54.down_proj.weight", "model.layers.19.mlp.experts.55.gate_and_up_proj.weight", "model.layers.19.mlp.experts.55.down_proj.weight", "model.layers.19.mlp.experts.56.gate_and_up_proj.weight", "model.layers.19.mlp.experts.56.down_proj.weight", "model.layers.19.mlp.experts.57.gate_and_up_proj.weight", "model.layers.19.mlp.experts.57.down_proj.weight", "model.layers.19.mlp.experts.58.gate_and_up_proj.weight", "model.layers.19.mlp.experts.58.down_proj.weight", "model.layers.19.mlp.experts.59.gate_and_up_proj.weight", "model.layers.19.mlp.experts.59.down_proj.weight", "model.layers.19.mlp.experts.60.gate_and_up_proj.weight", "model.layers.19.mlp.experts.60.down_proj.weight", "model.layers.19.mlp.experts.61.gate_and_up_proj.weight", "model.layers.19.mlp.experts.61.down_proj.weight", "model.layers.19.mlp.experts.62.gate_and_up_proj.weight", "model.layers.19.mlp.experts.62.down_proj.weight", "model.layers.19.mlp.experts.63.gate_and_up_proj.weight", "model.layers.19.mlp.experts.63.down_proj.weight", "model.layers.20.self_attn.qkv_proj.weight", "model.layers.20.self_attn.o_proj.weight", "model.layers.20.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.20.mlp.shared_mlp.down_proj.weight", "model.layers.20.mlp.experts.0.gate_and_up_proj.weight", "model.layers.20.mlp.experts.0.down_proj.weight", "model.layers.20.mlp.experts.1.gate_and_up_proj.weight", "model.layers.20.mlp.experts.1.down_proj.weight", "model.layers.20.mlp.experts.2.gate_and_up_proj.weight", "model.layers.20.mlp.experts.2.down_proj.weight", "model.layers.20.mlp.experts.3.gate_and_up_proj.weight", "model.layers.20.mlp.experts.3.down_proj.weight", "model.layers.20.mlp.experts.4.gate_and_up_proj.weight", "model.layers.20.mlp.experts.4.down_proj.weight", "model.layers.20.mlp.experts.5.gate_and_up_proj.weight", "model.layers.20.mlp.experts.5.down_proj.weight", "model.layers.20.mlp.experts.6.gate_and_up_proj.weight", "model.layers.20.mlp.experts.6.down_proj.weight", "model.layers.20.mlp.experts.7.gate_and_up_proj.weight", "model.layers.20.mlp.experts.7.down_proj.weight", "model.layers.20.mlp.experts.8.gate_and_up_proj.weight", "model.layers.20.mlp.experts.8.down_proj.weight", "model.layers.20.mlp.experts.9.gate_and_up_proj.weight", "model.layers.20.mlp.experts.9.down_proj.weight", "model.layers.20.mlp.experts.10.gate_and_up_proj.weight", "model.layers.20.mlp.experts.10.down_proj.weight", "model.layers.20.mlp.experts.11.gate_and_up_proj.weight", "model.layers.20.mlp.experts.11.down_proj.weight", "model.layers.20.mlp.experts.12.gate_and_up_proj.weight", "model.layers.20.mlp.experts.12.down_proj.weight", "model.layers.20.mlp.experts.13.gate_and_up_proj.weight", "model.layers.20.mlp.experts.13.down_proj.weight", "model.layers.20.mlp.experts.14.gate_and_up_proj.weight", "model.layers.20.mlp.experts.14.down_proj.weight", "model.layers.20.mlp.experts.15.gate_and_up_proj.weight", "model.layers.20.mlp.experts.15.down_proj.weight", "model.layers.20.mlp.experts.16.gate_and_up_proj.weight", "model.layers.20.mlp.experts.16.down_proj.weight", "model.layers.20.mlp.experts.17.gate_and_up_proj.weight", "model.layers.20.mlp.experts.17.down_proj.weight", "model.layers.20.mlp.experts.18.gate_and_up_proj.weight", "model.layers.20.mlp.experts.18.down_proj.weight", "model.layers.20.mlp.experts.19.gate_and_up_proj.weight", "model.layers.20.mlp.experts.19.down_proj.weight", "model.layers.20.mlp.experts.20.gate_and_up_proj.weight", "model.layers.20.mlp.experts.20.down_proj.weight", "model.layers.20.mlp.experts.21.gate_and_up_proj.weight", "model.layers.20.mlp.experts.21.down_proj.weight", "model.layers.20.mlp.experts.22.gate_and_up_proj.weight", "model.layers.20.mlp.experts.22.down_proj.weight", "model.layers.20.mlp.experts.23.gate_and_up_proj.weight", "model.layers.20.mlp.experts.23.down_proj.weight", "model.layers.20.mlp.experts.24.gate_and_up_proj.weight", "model.layers.20.mlp.experts.24.down_proj.weight", "model.layers.20.mlp.experts.25.gate_and_up_proj.weight", "model.layers.20.mlp.experts.25.down_proj.weight", "model.layers.20.mlp.experts.26.gate_and_up_proj.weight", "model.layers.20.mlp.experts.26.down_proj.weight", "model.layers.20.mlp.experts.27.gate_and_up_proj.weight", "model.layers.20.mlp.experts.27.down_proj.weight", "model.layers.20.mlp.experts.28.gate_and_up_proj.weight", "model.layers.20.mlp.experts.28.down_proj.weight", "model.layers.20.mlp.experts.29.gate_and_up_proj.weight", "model.layers.20.mlp.experts.29.down_proj.weight", "model.layers.20.mlp.experts.30.gate_and_up_proj.weight", "model.layers.20.mlp.experts.30.down_proj.weight", "model.layers.20.mlp.experts.31.gate_and_up_proj.weight", "model.layers.20.mlp.experts.31.down_proj.weight", "model.layers.20.mlp.experts.32.gate_and_up_proj.weight", "model.layers.20.mlp.experts.32.down_proj.weight", "model.layers.20.mlp.experts.33.gate_and_up_proj.weight", "model.layers.20.mlp.experts.33.down_proj.weight", "model.layers.20.mlp.experts.34.gate_and_up_proj.weight", "model.layers.20.mlp.experts.34.down_proj.weight", "model.layers.20.mlp.experts.35.gate_and_up_proj.weight", "model.layers.20.mlp.experts.35.down_proj.weight", "model.layers.20.mlp.experts.36.gate_and_up_proj.weight", "model.layers.20.mlp.experts.36.down_proj.weight", "model.layers.20.mlp.experts.37.gate_and_up_proj.weight", "model.layers.20.mlp.experts.37.down_proj.weight", "model.layers.20.mlp.experts.38.gate_and_up_proj.weight", "model.layers.20.mlp.experts.38.down_proj.weight", "model.layers.20.mlp.experts.39.gate_and_up_proj.weight", "model.layers.20.mlp.experts.39.down_proj.weight", "model.layers.20.mlp.experts.40.gate_and_up_proj.weight", "model.layers.20.mlp.experts.40.down_proj.weight", "model.layers.20.mlp.experts.41.gate_and_up_proj.weight", "model.layers.20.mlp.experts.41.down_proj.weight", "model.layers.20.mlp.experts.42.gate_and_up_proj.weight", "model.layers.20.mlp.experts.42.down_proj.weight", "model.layers.20.mlp.experts.43.gate_and_up_proj.weight", "model.layers.20.mlp.experts.43.down_proj.weight", "model.layers.20.mlp.experts.44.gate_and_up_proj.weight", "model.layers.20.mlp.experts.44.down_proj.weight", "model.layers.20.mlp.experts.45.gate_and_up_proj.weight", "model.layers.20.mlp.experts.45.down_proj.weight", "model.layers.20.mlp.experts.46.gate_and_up_proj.weight", "model.layers.20.mlp.experts.46.down_proj.weight", "model.layers.20.mlp.experts.47.gate_and_up_proj.weight", "model.layers.20.mlp.experts.47.down_proj.weight", "model.layers.20.mlp.experts.48.gate_and_up_proj.weight", "model.layers.20.mlp.experts.48.down_proj.weight", "model.layers.20.mlp.experts.49.gate_and_up_proj.weight", "model.layers.20.mlp.experts.49.down_proj.weight", "model.layers.20.mlp.experts.50.gate_and_up_proj.weight", "model.layers.20.mlp.experts.50.down_proj.weight", "model.layers.20.mlp.experts.51.gate_and_up_proj.weight", "model.layers.20.mlp.experts.51.down_proj.weight", "model.layers.20.mlp.experts.52.gate_and_up_proj.weight", "model.layers.20.mlp.experts.52.down_proj.weight", "model.layers.20.mlp.experts.53.gate_and_up_proj.weight", "model.layers.20.mlp.experts.53.down_proj.weight", "model.layers.20.mlp.experts.54.gate_and_up_proj.weight", "model.layers.20.mlp.experts.54.down_proj.weight", "model.layers.20.mlp.experts.55.gate_and_up_proj.weight", "model.layers.20.mlp.experts.55.down_proj.weight", "model.layers.20.mlp.experts.56.gate_and_up_proj.weight", "model.layers.20.mlp.experts.56.down_proj.weight", "model.layers.20.mlp.experts.57.gate_and_up_proj.weight", "model.layers.20.mlp.experts.57.down_proj.weight", "model.layers.20.mlp.experts.58.gate_and_up_proj.weight", "model.layers.20.mlp.experts.58.down_proj.weight", "model.layers.20.mlp.experts.59.gate_and_up_proj.weight", "model.layers.20.mlp.experts.59.down_proj.weight", "model.layers.20.mlp.experts.60.gate_and_up_proj.weight", "model.layers.20.mlp.experts.60.down_proj.weight", "model.layers.20.mlp.experts.61.gate_and_up_proj.weight", "model.layers.20.mlp.experts.61.down_proj.weight", "model.layers.20.mlp.experts.62.gate_and_up_proj.weight", "model.layers.20.mlp.experts.62.down_proj.weight", "model.layers.20.mlp.experts.63.gate_and_up_proj.weight", "model.layers.20.mlp.experts.63.down_proj.weight", "model.layers.21.self_attn.qkv_proj.weight", "model.layers.21.self_attn.o_proj.weight", "model.layers.21.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.21.mlp.shared_mlp.down_proj.weight", "model.layers.21.mlp.experts.0.gate_and_up_proj.weight", "model.layers.21.mlp.experts.0.down_proj.weight", "model.layers.21.mlp.experts.1.gate_and_up_proj.weight", "model.layers.21.mlp.experts.1.down_proj.weight", "model.layers.21.mlp.experts.2.gate_and_up_proj.weight", "model.layers.21.mlp.experts.2.down_proj.weight", "model.layers.21.mlp.experts.3.gate_and_up_proj.weight", "model.layers.21.mlp.experts.3.down_proj.weight", "model.layers.21.mlp.experts.4.gate_and_up_proj.weight", "model.layers.21.mlp.experts.4.down_proj.weight", "model.layers.21.mlp.experts.5.gate_and_up_proj.weight", "model.layers.21.mlp.experts.5.down_proj.weight", "model.layers.21.mlp.experts.6.gate_and_up_proj.weight", "model.layers.21.mlp.experts.6.down_proj.weight", "model.layers.21.mlp.experts.7.gate_and_up_proj.weight", "model.layers.21.mlp.experts.7.down_proj.weight", "model.layers.21.mlp.experts.8.gate_and_up_proj.weight", "model.layers.21.mlp.experts.8.down_proj.weight", "model.layers.21.mlp.experts.9.gate_and_up_proj.weight", "model.layers.21.mlp.experts.9.down_proj.weight", "model.layers.21.mlp.experts.10.gate_and_up_proj.weight", "model.layers.21.mlp.experts.10.down_proj.weight", "model.layers.21.mlp.experts.11.gate_and_up_proj.weight", "model.layers.21.mlp.experts.11.down_proj.weight", "model.layers.21.mlp.experts.12.gate_and_up_proj.weight", "model.layers.21.mlp.experts.12.down_proj.weight", "model.layers.21.mlp.experts.13.gate_and_up_proj.weight", "model.layers.21.mlp.experts.13.down_proj.weight", "model.layers.21.mlp.experts.14.gate_and_up_proj.weight", "model.layers.21.mlp.experts.14.down_proj.weight", "model.layers.21.mlp.experts.15.gate_and_up_proj.weight", "model.layers.21.mlp.experts.15.down_proj.weight", "model.layers.21.mlp.experts.16.gate_and_up_proj.weight", "model.layers.21.mlp.experts.16.down_proj.weight", "model.layers.21.mlp.experts.17.gate_and_up_proj.weight", "model.layers.21.mlp.experts.17.down_proj.weight", "model.layers.21.mlp.experts.18.gate_and_up_proj.weight", "model.layers.21.mlp.experts.18.down_proj.weight", "model.layers.21.mlp.experts.19.gate_and_up_proj.weight", "model.layers.21.mlp.experts.19.down_proj.weight", "model.layers.21.mlp.experts.20.gate_and_up_proj.weight", "model.layers.21.mlp.experts.20.down_proj.weight", "model.layers.21.mlp.experts.21.gate_and_up_proj.weight", "model.layers.21.mlp.experts.21.down_proj.weight", "model.layers.21.mlp.experts.22.gate_and_up_proj.weight", "model.layers.21.mlp.experts.22.down_proj.weight", "model.layers.21.mlp.experts.23.gate_and_up_proj.weight", "model.layers.21.mlp.experts.23.down_proj.weight", "model.layers.21.mlp.experts.24.gate_and_up_proj.weight", "model.layers.21.mlp.experts.24.down_proj.weight", "model.layers.21.mlp.experts.25.gate_and_up_proj.weight", "model.layers.21.mlp.experts.25.down_proj.weight", "model.layers.21.mlp.experts.26.gate_and_up_proj.weight", "model.layers.21.mlp.experts.26.down_proj.weight", "model.layers.21.mlp.experts.27.gate_and_up_proj.weight", "model.layers.21.mlp.experts.27.down_proj.weight", "model.layers.21.mlp.experts.28.gate_and_up_proj.weight", "model.layers.21.mlp.experts.28.down_proj.weight", "model.layers.21.mlp.experts.29.gate_and_up_proj.weight", "model.layers.21.mlp.experts.29.down_proj.weight", "model.layers.21.mlp.experts.30.gate_and_up_proj.weight", "model.layers.21.mlp.experts.30.down_proj.weight", "model.layers.21.mlp.experts.31.gate_and_up_proj.weight", "model.layers.21.mlp.experts.31.down_proj.weight", "model.layers.21.mlp.experts.32.gate_and_up_proj.weight", "model.layers.21.mlp.experts.32.down_proj.weight", "model.layers.21.mlp.experts.33.gate_and_up_proj.weight", "model.layers.21.mlp.experts.33.down_proj.weight", "model.layers.21.mlp.experts.34.gate_and_up_proj.weight", "model.layers.21.mlp.experts.34.down_proj.weight", "model.layers.21.mlp.experts.35.gate_and_up_proj.weight", "model.layers.21.mlp.experts.35.down_proj.weight", "model.layers.21.mlp.experts.36.gate_and_up_proj.weight", "model.layers.21.mlp.experts.36.down_proj.weight", "model.layers.21.mlp.experts.37.gate_and_up_proj.weight", "model.layers.21.mlp.experts.37.down_proj.weight", "model.layers.21.mlp.experts.38.gate_and_up_proj.weight", "model.layers.21.mlp.experts.38.down_proj.weight", "model.layers.21.mlp.experts.39.gate_and_up_proj.weight", "model.layers.21.mlp.experts.39.down_proj.weight", "model.layers.21.mlp.experts.40.gate_and_up_proj.weight", "model.layers.21.mlp.experts.40.down_proj.weight", "model.layers.21.mlp.experts.41.gate_and_up_proj.weight", "model.layers.21.mlp.experts.41.down_proj.weight", "model.layers.21.mlp.experts.42.gate_and_up_proj.weight", "model.layers.21.mlp.experts.42.down_proj.weight", "model.layers.21.mlp.experts.43.gate_and_up_proj.weight", "model.layers.21.mlp.experts.43.down_proj.weight", "model.layers.21.mlp.experts.44.gate_and_up_proj.weight", "model.layers.21.mlp.experts.44.down_proj.weight", "model.layers.21.mlp.experts.45.gate_and_up_proj.weight", "model.layers.21.mlp.experts.45.down_proj.weight", "model.layers.21.mlp.experts.46.gate_and_up_proj.weight", "model.layers.21.mlp.experts.46.down_proj.weight", "model.layers.21.mlp.experts.47.gate_and_up_proj.weight", "model.layers.21.mlp.experts.47.down_proj.weight", "model.layers.21.mlp.experts.48.gate_and_up_proj.weight", "model.layers.21.mlp.experts.48.down_proj.weight", "model.layers.21.mlp.experts.49.gate_and_up_proj.weight", "model.layers.21.mlp.experts.49.down_proj.weight", "model.layers.21.mlp.experts.50.gate_and_up_proj.weight", "model.layers.21.mlp.experts.50.down_proj.weight", "model.layers.21.mlp.experts.51.gate_and_up_proj.weight", "model.layers.21.mlp.experts.51.down_proj.weight", "model.layers.21.mlp.experts.52.gate_and_up_proj.weight", "model.layers.21.mlp.experts.52.down_proj.weight", "model.layers.21.mlp.experts.53.gate_and_up_proj.weight", "model.layers.21.mlp.experts.53.down_proj.weight", "model.layers.21.mlp.experts.54.gate_and_up_proj.weight", "model.layers.21.mlp.experts.54.down_proj.weight", "model.layers.21.mlp.experts.55.gate_and_up_proj.weight", "model.layers.21.mlp.experts.55.down_proj.weight", "model.layers.21.mlp.experts.56.gate_and_up_proj.weight", "model.layers.21.mlp.experts.56.down_proj.weight", "model.layers.21.mlp.experts.57.gate_and_up_proj.weight", "model.layers.21.mlp.experts.57.down_proj.weight", "model.layers.21.mlp.experts.58.gate_and_up_proj.weight", "model.layers.21.mlp.experts.58.down_proj.weight", "model.layers.21.mlp.experts.59.gate_and_up_proj.weight", "model.layers.21.mlp.experts.59.down_proj.weight", "model.layers.21.mlp.experts.60.gate_and_up_proj.weight", "model.layers.21.mlp.experts.60.down_proj.weight", "model.layers.21.mlp.experts.61.gate_and_up_proj.weight", "model.layers.21.mlp.experts.61.down_proj.weight", "model.layers.21.mlp.experts.62.gate_and_up_proj.weight", "model.layers.21.mlp.experts.62.down_proj.weight", "model.layers.21.mlp.experts.63.gate_and_up_proj.weight", "model.layers.21.mlp.experts.63.down_proj.weight", "model.layers.22.self_attn.qkv_proj.weight", "model.layers.22.self_attn.o_proj.weight", "model.layers.22.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.22.mlp.shared_mlp.down_proj.weight", "model.layers.22.mlp.experts.0.gate_and_up_proj.weight", "model.layers.22.mlp.experts.0.down_proj.weight", "model.layers.22.mlp.experts.1.gate_and_up_proj.weight", "model.layers.22.mlp.experts.1.down_proj.weight", "model.layers.22.mlp.experts.2.gate_and_up_proj.weight", "model.layers.22.mlp.experts.2.down_proj.weight", "model.layers.22.mlp.experts.3.gate_and_up_proj.weight", "model.layers.22.mlp.experts.3.down_proj.weight", "model.layers.22.mlp.experts.4.gate_and_up_proj.weight", "model.layers.22.mlp.experts.4.down_proj.weight", "model.layers.22.mlp.experts.5.gate_and_up_proj.weight", "model.layers.22.mlp.experts.5.down_proj.weight", "model.layers.22.mlp.experts.6.gate_and_up_proj.weight", "model.layers.22.mlp.experts.6.down_proj.weight", "model.layers.22.mlp.experts.7.gate_and_up_proj.weight", "model.layers.22.mlp.experts.7.down_proj.weight", "model.layers.22.mlp.experts.8.gate_and_up_proj.weight", "model.layers.22.mlp.experts.8.down_proj.weight", "model.layers.22.mlp.experts.9.gate_and_up_proj.weight", "model.layers.22.mlp.experts.9.down_proj.weight", "model.layers.22.mlp.experts.10.gate_and_up_proj.weight", "model.layers.22.mlp.experts.10.down_proj.weight", "model.layers.22.mlp.experts.11.gate_and_up_proj.weight", "model.layers.22.mlp.experts.11.down_proj.weight", "model.layers.22.mlp.experts.12.gate_and_up_proj.weight", "model.layers.22.mlp.experts.12.down_proj.weight", "model.layers.22.mlp.experts.13.gate_and_up_proj.weight", "model.layers.22.mlp.experts.13.down_proj.weight", "model.layers.22.mlp.experts.14.gate_and_up_proj.weight", "model.layers.22.mlp.experts.14.down_proj.weight", "model.layers.22.mlp.experts.15.gate_and_up_proj.weight", "model.layers.22.mlp.experts.15.down_proj.weight", "model.layers.22.mlp.experts.16.gate_and_up_proj.weight", "model.layers.22.mlp.experts.16.down_proj.weight", "model.layers.22.mlp.experts.17.gate_and_up_proj.weight", "model.layers.22.mlp.experts.17.down_proj.weight", "model.layers.22.mlp.experts.18.gate_and_up_proj.weight", "model.layers.22.mlp.experts.18.down_proj.weight", "model.layers.22.mlp.experts.19.gate_and_up_proj.weight", "model.layers.22.mlp.experts.19.down_proj.weight", "model.layers.22.mlp.experts.20.gate_and_up_proj.weight", "model.layers.22.mlp.experts.20.down_proj.weight", "model.layers.22.mlp.experts.21.gate_and_up_proj.weight", "model.layers.22.mlp.experts.21.down_proj.weight", "model.layers.22.mlp.experts.22.gate_and_up_proj.weight", "model.layers.22.mlp.experts.22.down_proj.weight", "model.layers.22.mlp.experts.23.gate_and_up_proj.weight", "model.layers.22.mlp.experts.23.down_proj.weight", "model.layers.22.mlp.experts.24.gate_and_up_proj.weight", "model.layers.22.mlp.experts.24.down_proj.weight", "model.layers.22.mlp.experts.25.gate_and_up_proj.weight", "model.layers.22.mlp.experts.25.down_proj.weight", "model.layers.22.mlp.experts.26.gate_and_up_proj.weight", "model.layers.22.mlp.experts.26.down_proj.weight", "model.layers.22.mlp.experts.27.gate_and_up_proj.weight", "model.layers.22.mlp.experts.27.down_proj.weight", "model.layers.22.mlp.experts.28.gate_and_up_proj.weight", "model.layers.22.mlp.experts.28.down_proj.weight", "model.layers.22.mlp.experts.29.gate_and_up_proj.weight", "model.layers.22.mlp.experts.29.down_proj.weight", "model.layers.22.mlp.experts.30.gate_and_up_proj.weight", "model.layers.22.mlp.experts.30.down_proj.weight", "model.layers.22.mlp.experts.31.gate_and_up_proj.weight", "model.layers.22.mlp.experts.31.down_proj.weight", "model.layers.22.mlp.experts.32.gate_and_up_proj.weight", "model.layers.22.mlp.experts.32.down_proj.weight", "model.layers.22.mlp.experts.33.gate_and_up_proj.weight", "model.layers.22.mlp.experts.33.down_proj.weight", "model.layers.22.mlp.experts.34.gate_and_up_proj.weight", "model.layers.22.mlp.experts.34.down_proj.weight", "model.layers.22.mlp.experts.35.gate_and_up_proj.weight", "model.layers.22.mlp.experts.35.down_proj.weight", "model.layers.22.mlp.experts.36.gate_and_up_proj.weight", "model.layers.22.mlp.experts.36.down_proj.weight", "model.layers.22.mlp.experts.37.gate_and_up_proj.weight", "model.layers.22.mlp.experts.37.down_proj.weight", "model.layers.22.mlp.experts.38.gate_and_up_proj.weight", "model.layers.22.mlp.experts.38.down_proj.weight", "model.layers.22.mlp.experts.39.gate_and_up_proj.weight", "model.layers.22.mlp.experts.39.down_proj.weight", "model.layers.22.mlp.experts.40.gate_and_up_proj.weight", "model.layers.22.mlp.experts.40.down_proj.weight", "model.layers.22.mlp.experts.41.gate_and_up_proj.weight", "model.layers.22.mlp.experts.41.down_proj.weight", "model.layers.22.mlp.experts.42.gate_and_up_proj.weight", "model.layers.22.mlp.experts.42.down_proj.weight", "model.layers.22.mlp.experts.43.gate_and_up_proj.weight", "model.layers.22.mlp.experts.43.down_proj.weight", "model.layers.22.mlp.experts.44.gate_and_up_proj.weight", "model.layers.22.mlp.experts.44.down_proj.weight", "model.layers.22.mlp.experts.45.gate_and_up_proj.weight", "model.layers.22.mlp.experts.45.down_proj.weight", "model.layers.22.mlp.experts.46.gate_and_up_proj.weight", "model.layers.22.mlp.experts.46.down_proj.weight", "model.layers.22.mlp.experts.47.gate_and_up_proj.weight", "model.layers.22.mlp.experts.47.down_proj.weight", "model.layers.22.mlp.experts.48.gate_and_up_proj.weight", "model.layers.22.mlp.experts.48.down_proj.weight", "model.layers.22.mlp.experts.49.gate_and_up_proj.weight", "model.layers.22.mlp.experts.49.down_proj.weight", "model.layers.22.mlp.experts.50.gate_and_up_proj.weight", "model.layers.22.mlp.experts.50.down_proj.weight", "model.layers.22.mlp.experts.51.gate_and_up_proj.weight", "model.layers.22.mlp.experts.51.down_proj.weight", "model.layers.22.mlp.experts.52.gate_and_up_proj.weight", "model.layers.22.mlp.experts.52.down_proj.weight", "model.layers.22.mlp.experts.53.gate_and_up_proj.weight", "model.layers.22.mlp.experts.53.down_proj.weight", "model.layers.22.mlp.experts.54.gate_and_up_proj.weight", "model.layers.22.mlp.experts.54.down_proj.weight", "model.layers.22.mlp.experts.55.gate_and_up_proj.weight", "model.layers.22.mlp.experts.55.down_proj.weight", "model.layers.22.mlp.experts.56.gate_and_up_proj.weight", "model.layers.22.mlp.experts.56.down_proj.weight", "model.layers.22.mlp.experts.57.gate_and_up_proj.weight", "model.layers.22.mlp.experts.57.down_proj.weight", "model.layers.22.mlp.experts.58.gate_and_up_proj.weight", "model.layers.22.mlp.experts.58.down_proj.weight", "model.layers.22.mlp.experts.59.gate_and_up_proj.weight", "model.layers.22.mlp.experts.59.down_proj.weight", "model.layers.22.mlp.experts.60.gate_and_up_proj.weight", "model.layers.22.mlp.experts.60.down_proj.weight", "model.layers.22.mlp.experts.61.gate_and_up_proj.weight", "model.layers.22.mlp.experts.61.down_proj.weight", "model.layers.22.mlp.experts.62.gate_and_up_proj.weight", "model.layers.22.mlp.experts.62.down_proj.weight", "model.layers.22.mlp.experts.63.gate_and_up_proj.weight", "model.layers.22.mlp.experts.63.down_proj.weight", "model.layers.23.self_attn.qkv_proj.weight", "model.layers.23.self_attn.o_proj.weight", "model.layers.23.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.23.mlp.shared_mlp.down_proj.weight", "model.layers.23.mlp.experts.0.gate_and_up_proj.weight", "model.layers.23.mlp.experts.0.down_proj.weight", "model.layers.23.mlp.experts.1.gate_and_up_proj.weight", "model.layers.23.mlp.experts.1.down_proj.weight", "model.layers.23.mlp.experts.2.gate_and_up_proj.weight", "model.layers.23.mlp.experts.2.down_proj.weight", "model.layers.23.mlp.experts.3.gate_and_up_proj.weight", "model.layers.23.mlp.experts.3.down_proj.weight", "model.layers.23.mlp.experts.4.gate_and_up_proj.weight", "model.layers.23.mlp.experts.4.down_proj.weight", "model.layers.23.mlp.experts.5.gate_and_up_proj.weight", "model.layers.23.mlp.experts.5.down_proj.weight", "model.layers.23.mlp.experts.6.gate_and_up_proj.weight", "model.layers.23.mlp.experts.6.down_proj.weight", "model.layers.23.mlp.experts.7.gate_and_up_proj.weight", "model.layers.23.mlp.experts.7.down_proj.weight", "model.layers.23.mlp.experts.8.gate_and_up_proj.weight", "model.layers.23.mlp.experts.8.down_proj.weight", "model.layers.23.mlp.experts.9.gate_and_up_proj.weight", "model.layers.23.mlp.experts.9.down_proj.weight", "model.layers.23.mlp.experts.10.gate_and_up_proj.weight", "model.layers.23.mlp.experts.10.down_proj.weight", "model.layers.23.mlp.experts.11.gate_and_up_proj.weight", "model.layers.23.mlp.experts.11.down_proj.weight", "model.layers.23.mlp.experts.12.gate_and_up_proj.weight", "model.layers.23.mlp.experts.12.down_proj.weight", "model.layers.23.mlp.experts.13.gate_and_up_proj.weight", "model.layers.23.mlp.experts.13.down_proj.weight", "model.layers.23.mlp.experts.14.gate_and_up_proj.weight", "model.layers.23.mlp.experts.14.down_proj.weight", "model.layers.23.mlp.experts.15.gate_and_up_proj.weight", "model.layers.23.mlp.experts.15.down_proj.weight", "model.layers.23.mlp.experts.16.gate_and_up_proj.weight", "model.layers.23.mlp.experts.16.down_proj.weight", "model.layers.23.mlp.experts.17.gate_and_up_proj.weight", "model.layers.23.mlp.experts.17.down_proj.weight", "model.layers.23.mlp.experts.18.gate_and_up_proj.weight", "model.layers.23.mlp.experts.18.down_proj.weight", "model.layers.23.mlp.experts.19.gate_and_up_proj.weight", "model.layers.23.mlp.experts.19.down_proj.weight", "model.layers.23.mlp.experts.20.gate_and_up_proj.weight", "model.layers.23.mlp.experts.20.down_proj.weight", "model.layers.23.mlp.experts.21.gate_and_up_proj.weight", "model.layers.23.mlp.experts.21.down_proj.weight", "model.layers.23.mlp.experts.22.gate_and_up_proj.weight", "model.layers.23.mlp.experts.22.down_proj.weight", "model.layers.23.mlp.experts.23.gate_and_up_proj.weight", "model.layers.23.mlp.experts.23.down_proj.weight", "model.layers.23.mlp.experts.24.gate_and_up_proj.weight", "model.layers.23.mlp.experts.24.down_proj.weight", "model.layers.23.mlp.experts.25.gate_and_up_proj.weight", "model.layers.23.mlp.experts.25.down_proj.weight", "model.layers.23.mlp.experts.26.gate_and_up_proj.weight", "model.layers.23.mlp.experts.26.down_proj.weight", "model.layers.23.mlp.experts.27.gate_and_up_proj.weight", "model.layers.23.mlp.experts.27.down_proj.weight", "model.layers.23.mlp.experts.28.gate_and_up_proj.weight", "model.layers.23.mlp.experts.28.down_proj.weight", "model.layers.23.mlp.experts.29.gate_and_up_proj.weight", "model.layers.23.mlp.experts.29.down_proj.weight", "model.layers.23.mlp.experts.30.gate_and_up_proj.weight", "model.layers.23.mlp.experts.30.down_proj.weight", "model.layers.23.mlp.experts.31.gate_and_up_proj.weight", "model.layers.23.mlp.experts.31.down_proj.weight", "model.layers.23.mlp.experts.32.gate_and_up_proj.weight", "model.layers.23.mlp.experts.32.down_proj.weight", "model.layers.23.mlp.experts.33.gate_and_up_proj.weight", "model.layers.23.mlp.experts.33.down_proj.weight", "model.layers.23.mlp.experts.34.gate_and_up_proj.weight", "model.layers.23.mlp.experts.34.down_proj.weight", "model.layers.23.mlp.experts.35.gate_and_up_proj.weight", "model.layers.23.mlp.experts.35.down_proj.weight", "model.layers.23.mlp.experts.36.gate_and_up_proj.weight", "model.layers.23.mlp.experts.36.down_proj.weight", "model.layers.23.mlp.experts.37.gate_and_up_proj.weight", "model.layers.23.mlp.experts.37.down_proj.weight", "model.layers.23.mlp.experts.38.gate_and_up_proj.weight", "model.layers.23.mlp.experts.38.down_proj.weight", "model.layers.23.mlp.experts.39.gate_and_up_proj.weight", "model.layers.23.mlp.experts.39.down_proj.weight", "model.layers.23.mlp.experts.40.gate_and_up_proj.weight", "model.layers.23.mlp.experts.40.down_proj.weight", "model.layers.23.mlp.experts.41.gate_and_up_proj.weight", "model.layers.23.mlp.experts.41.down_proj.weight", "model.layers.23.mlp.experts.42.gate_and_up_proj.weight", "model.layers.23.mlp.experts.42.down_proj.weight", "model.layers.23.mlp.experts.43.gate_and_up_proj.weight", "model.layers.23.mlp.experts.43.down_proj.weight", "model.layers.23.mlp.experts.44.gate_and_up_proj.weight", "model.layers.23.mlp.experts.44.down_proj.weight", "model.layers.23.mlp.experts.45.gate_and_up_proj.weight", "model.layers.23.mlp.experts.45.down_proj.weight", "model.layers.23.mlp.experts.46.gate_and_up_proj.weight", "model.layers.23.mlp.experts.46.down_proj.weight", "model.layers.23.mlp.experts.47.gate_and_up_proj.weight", "model.layers.23.mlp.experts.47.down_proj.weight", "model.layers.23.mlp.experts.48.gate_and_up_proj.weight", "model.layers.23.mlp.experts.48.down_proj.weight", "model.layers.23.mlp.experts.49.gate_and_up_proj.weight", "model.layers.23.mlp.experts.49.down_proj.weight", "model.layers.23.mlp.experts.50.gate_and_up_proj.weight", "model.layers.23.mlp.experts.50.down_proj.weight", "model.layers.23.mlp.experts.51.gate_and_up_proj.weight", "model.layers.23.mlp.experts.51.down_proj.weight", "model.layers.23.mlp.experts.52.gate_and_up_proj.weight", "model.layers.23.mlp.experts.52.down_proj.weight", "model.layers.23.mlp.experts.53.gate_and_up_proj.weight", "model.layers.23.mlp.experts.53.down_proj.weight", "model.layers.23.mlp.experts.54.gate_and_up_proj.weight", "model.layers.23.mlp.experts.54.down_proj.weight", "model.layers.23.mlp.experts.55.gate_and_up_proj.weight", "model.layers.23.mlp.experts.55.down_proj.weight", "model.layers.23.mlp.experts.56.gate_and_up_proj.weight", "model.layers.23.mlp.experts.56.down_proj.weight", "model.layers.23.mlp.experts.57.gate_and_up_proj.weight", "model.layers.23.mlp.experts.57.down_proj.weight", "model.layers.23.mlp.experts.58.gate_and_up_proj.weight", "model.layers.23.mlp.experts.58.down_proj.weight", "model.layers.23.mlp.experts.59.gate_and_up_proj.weight", "model.layers.23.mlp.experts.59.down_proj.weight", "model.layers.23.mlp.experts.60.gate_and_up_proj.weight", "model.layers.23.mlp.experts.60.down_proj.weight", "model.layers.23.mlp.experts.61.gate_and_up_proj.weight", "model.layers.23.mlp.experts.61.down_proj.weight", "model.layers.23.mlp.experts.62.gate_and_up_proj.weight", "model.layers.23.mlp.experts.62.down_proj.weight", "model.layers.23.mlp.experts.63.gate_and_up_proj.weight", "model.layers.23.mlp.experts.63.down_proj.weight", "model.layers.24.self_attn.qkv_proj.weight", "model.layers.24.self_attn.o_proj.weight", "model.layers.24.mlp.shared_mlp.down_proj.weight", "model.layers.24.mlp.experts.0.gate_and_up_proj.weight", "model.layers.24.mlp.experts.0.down_proj.weight", "model.layers.24.mlp.experts.1.gate_and_up_proj.weight", "model.layers.24.mlp.experts.1.down_proj.weight", "model.layers.24.mlp.experts.2.gate_and_up_proj.weight", "model.layers.24.mlp.experts.2.down_proj.weight", "model.layers.24.mlp.experts.3.gate_and_up_proj.weight", "model.layers.24.mlp.experts.3.down_proj.weight", "model.layers.24.mlp.experts.4.gate_and_up_proj.weight", "model.layers.24.mlp.experts.4.down_proj.weight", "model.layers.24.mlp.experts.5.gate_and_up_proj.weight", "model.layers.24.mlp.experts.5.down_proj.weight", "model.layers.24.mlp.experts.6.gate_and_up_proj.weight", "model.layers.24.mlp.experts.6.down_proj.weight", "model.layers.24.mlp.experts.7.gate_and_up_proj.weight", "model.layers.24.mlp.experts.7.down_proj.weight", "model.layers.24.mlp.experts.8.gate_and_up_proj.weight", "model.layers.24.mlp.experts.8.down_proj.weight", "model.layers.24.mlp.experts.9.gate_and_up_proj.weight", "model.layers.24.mlp.experts.9.down_proj.weight", "model.layers.24.mlp.experts.10.gate_and_up_proj.weight", "model.layers.24.mlp.experts.10.down_proj.weight", "model.layers.24.mlp.experts.11.gate_and_up_proj.weight", "model.layers.24.mlp.experts.11.down_proj.weight", "model.layers.24.mlp.experts.12.gate_and_up_proj.weight", "model.layers.24.mlp.experts.12.down_proj.weight", "model.layers.24.mlp.experts.13.gate_and_up_proj.weight", "model.layers.24.mlp.experts.13.down_proj.weight", "model.layers.24.mlp.experts.14.gate_and_up_proj.weight", "model.layers.24.mlp.experts.14.down_proj.weight", "model.layers.24.mlp.experts.15.gate_and_up_proj.weight", "model.layers.24.mlp.experts.15.down_proj.weight", "model.layers.24.mlp.experts.16.gate_and_up_proj.weight", "model.layers.24.mlp.experts.16.down_proj.weight", "model.layers.24.mlp.experts.17.gate_and_up_proj.weight", "model.layers.24.mlp.experts.17.down_proj.weight", "model.layers.24.mlp.experts.18.gate_and_up_proj.weight", "model.layers.24.mlp.experts.18.down_proj.weight", "model.layers.24.mlp.experts.19.gate_and_up_proj.weight", "model.layers.24.mlp.experts.19.down_proj.weight", "model.layers.24.mlp.experts.20.gate_and_up_proj.weight", "model.layers.24.mlp.experts.20.down_proj.weight", "model.layers.24.mlp.experts.21.gate_and_up_proj.weight", "model.layers.24.mlp.experts.21.down_proj.weight", "model.layers.24.mlp.experts.22.gate_and_up_proj.weight", "model.layers.24.mlp.experts.22.down_proj.weight", "model.layers.24.mlp.experts.23.gate_and_up_proj.weight", "model.layers.24.mlp.experts.23.down_proj.weight", "model.layers.24.mlp.experts.24.gate_and_up_proj.weight", "model.layers.24.mlp.experts.24.down_proj.weight", "model.layers.24.mlp.experts.25.gate_and_up_proj.weight", "model.layers.24.mlp.experts.25.down_proj.weight", "model.layers.24.mlp.experts.26.gate_and_up_proj.weight", "model.layers.24.mlp.experts.26.down_proj.weight", "model.layers.24.mlp.experts.27.gate_and_up_proj.weight", "model.layers.24.mlp.experts.27.down_proj.weight", "model.layers.24.mlp.experts.28.gate_and_up_proj.weight", "model.layers.24.mlp.experts.28.down_proj.weight", "model.layers.24.mlp.experts.29.gate_and_up_proj.weight", "model.layers.24.mlp.experts.29.down_proj.weight", "model.layers.24.mlp.experts.30.gate_and_up_proj.weight", "model.layers.24.mlp.experts.30.down_proj.weight", "model.layers.24.mlp.experts.31.gate_and_up_proj.weight", "model.layers.24.mlp.experts.31.down_proj.weight", "model.layers.24.mlp.experts.32.gate_and_up_proj.weight", "model.layers.24.mlp.experts.32.down_proj.weight", "model.layers.24.mlp.experts.33.gate_and_up_proj.weight", "model.layers.24.mlp.experts.33.down_proj.weight", "model.layers.24.mlp.experts.34.gate_and_up_proj.weight", "model.layers.24.mlp.experts.34.down_proj.weight", "model.layers.24.mlp.experts.35.gate_and_up_proj.weight", "model.layers.24.mlp.experts.35.down_proj.weight", "model.layers.24.mlp.experts.36.gate_and_up_proj.weight", "model.layers.24.mlp.experts.36.down_proj.weight", "model.layers.24.mlp.experts.37.gate_and_up_proj.weight", "model.layers.24.mlp.experts.37.down_proj.weight", "model.layers.24.mlp.experts.38.gate_and_up_proj.weight", "model.layers.24.mlp.experts.38.down_proj.weight", "model.layers.24.mlp.experts.39.gate_and_up_proj.weight", "model.layers.24.mlp.experts.39.down_proj.weight", "model.layers.24.mlp.experts.40.gate_and_up_proj.weight", "model.layers.24.mlp.experts.40.down_proj.weight", "model.layers.24.mlp.experts.41.gate_and_up_proj.weight", "model.layers.24.mlp.experts.41.down_proj.weight", "model.layers.24.mlp.experts.42.gate_and_up_proj.weight", "model.layers.24.mlp.experts.42.down_proj.weight", "model.layers.24.mlp.experts.43.gate_and_up_proj.weight", "model.layers.24.mlp.experts.43.down_proj.weight", "model.layers.24.mlp.experts.44.gate_and_up_proj.weight", "model.layers.24.mlp.experts.44.down_proj.weight", "model.layers.24.mlp.experts.45.gate_and_up_proj.weight", "model.layers.24.mlp.experts.45.down_proj.weight", "model.layers.24.mlp.experts.46.gate_and_up_proj.weight", "model.layers.24.mlp.experts.46.down_proj.weight", "model.layers.24.mlp.experts.47.gate_and_up_proj.weight", "model.layers.24.mlp.experts.47.down_proj.weight", "model.layers.24.mlp.experts.48.gate_and_up_proj.weight", "model.layers.24.mlp.experts.48.down_proj.weight", "model.layers.24.mlp.experts.49.gate_and_up_proj.weight", "model.layers.24.mlp.experts.49.down_proj.weight", "model.layers.24.mlp.experts.50.gate_and_up_proj.weight", "model.layers.24.mlp.experts.50.down_proj.weight", "model.layers.24.mlp.experts.51.gate_and_up_proj.weight", "model.layers.24.mlp.experts.51.down_proj.weight", "model.layers.24.mlp.experts.52.gate_and_up_proj.weight", "model.layers.24.mlp.experts.52.down_proj.weight", "model.layers.24.mlp.experts.53.gate_and_up_proj.weight", "model.layers.24.mlp.experts.53.down_proj.weight", "model.layers.24.mlp.experts.54.gate_and_up_proj.weight", "model.layers.24.mlp.experts.54.down_proj.weight", "model.layers.24.mlp.experts.55.gate_and_up_proj.weight", "model.layers.24.mlp.experts.55.down_proj.weight", "model.layers.24.mlp.experts.56.gate_and_up_proj.weight", "model.layers.24.mlp.experts.56.down_proj.weight", "model.layers.24.mlp.experts.57.gate_and_up_proj.weight", "model.layers.24.mlp.experts.57.down_proj.weight", "model.layers.24.mlp.experts.58.gate_and_up_proj.weight", "model.layers.24.mlp.experts.58.down_proj.weight", "model.layers.24.mlp.experts.59.gate_and_up_proj.weight", "model.layers.24.mlp.experts.59.down_proj.weight", "model.layers.24.mlp.experts.60.gate_and_up_proj.weight", "model.layers.24.mlp.experts.60.down_proj.weight", "model.layers.24.mlp.experts.61.gate_and_up_proj.weight", "model.layers.24.mlp.experts.61.down_proj.weight", "model.layers.24.mlp.experts.62.gate_and_up_proj.weight", "model.layers.24.mlp.experts.62.down_proj.weight", "model.layers.24.mlp.experts.63.gate_and_up_proj.weight", "model.layers.24.mlp.experts.63.down_proj.weight", "model.layers.25.self_attn.qkv_proj.weight", "model.layers.25.self_attn.o_proj.weight", "model.layers.25.mlp.shared_mlp.down_proj.weight", "model.layers.25.mlp.experts.0.gate_and_up_proj.weight", "model.layers.25.mlp.experts.0.down_proj.weight", "model.layers.25.mlp.experts.1.gate_and_up_proj.weight", "model.layers.25.mlp.experts.1.down_proj.weight", "model.layers.25.mlp.experts.2.gate_and_up_proj.weight", "model.layers.25.mlp.experts.2.down_proj.weight", "model.layers.25.mlp.experts.3.gate_and_up_proj.weight", "model.layers.25.mlp.experts.3.down_proj.weight", "model.layers.25.mlp.experts.4.gate_and_up_proj.weight", "model.layers.25.mlp.experts.4.down_proj.weight", "model.layers.25.mlp.experts.5.gate_and_up_proj.weight", "model.layers.25.mlp.experts.5.down_proj.weight", "model.layers.25.mlp.experts.6.gate_and_up_proj.weight", "model.layers.25.mlp.experts.6.down_proj.weight", "model.layers.25.mlp.experts.7.gate_and_up_proj.weight", "model.layers.25.mlp.experts.7.down_proj.weight", "model.layers.25.mlp.experts.8.gate_and_up_proj.weight", "model.layers.25.mlp.experts.8.down_proj.weight", "model.layers.25.mlp.experts.9.gate_and_up_proj.weight", "model.layers.25.mlp.experts.9.down_proj.weight", "model.layers.25.mlp.experts.10.gate_and_up_proj.weight", "model.layers.25.mlp.experts.10.down_proj.weight", "model.layers.25.mlp.experts.11.gate_and_up_proj.weight", "model.layers.25.mlp.experts.11.down_proj.weight", "model.layers.25.mlp.experts.12.gate_and_up_proj.weight", "model.layers.25.mlp.experts.12.down_proj.weight", "model.layers.25.mlp.experts.13.gate_and_up_proj.weight", "model.layers.25.mlp.experts.13.down_proj.weight", "model.layers.25.mlp.experts.14.gate_and_up_proj.weight", "model.layers.25.mlp.experts.14.down_proj.weight", "model.layers.25.mlp.experts.15.gate_and_up_proj.weight", "model.layers.25.mlp.experts.15.down_proj.weight", "model.layers.25.mlp.experts.16.gate_and_up_proj.weight", "model.layers.25.mlp.experts.16.down_proj.weight", "model.layers.25.mlp.experts.17.gate_and_up_proj.weight", "model.layers.25.mlp.experts.17.down_proj.weight", "model.layers.25.mlp.experts.18.gate_and_up_proj.weight", "model.layers.25.mlp.experts.18.down_proj.weight", "model.layers.25.mlp.experts.19.gate_and_up_proj.weight", "model.layers.25.mlp.experts.19.down_proj.weight", "model.layers.25.mlp.experts.20.gate_and_up_proj.weight", "model.layers.25.mlp.experts.20.down_proj.weight", "model.layers.25.mlp.experts.21.gate_and_up_proj.weight", "model.layers.25.mlp.experts.21.down_proj.weight", "model.layers.25.mlp.experts.22.gate_and_up_proj.weight", "model.layers.25.mlp.experts.22.down_proj.weight", "model.layers.25.mlp.experts.23.gate_and_up_proj.weight", "model.layers.25.mlp.experts.23.down_proj.weight", "model.layers.25.mlp.experts.24.gate_and_up_proj.weight", "model.layers.25.mlp.experts.24.down_proj.weight", "model.layers.25.mlp.experts.25.gate_and_up_proj.weight", "model.layers.25.mlp.experts.25.down_proj.weight", "model.layers.25.mlp.experts.26.gate_and_up_proj.weight", "model.layers.25.mlp.experts.26.down_proj.weight", "model.layers.25.mlp.experts.27.gate_and_up_proj.weight", "model.layers.25.mlp.experts.27.down_proj.weight", "model.layers.25.mlp.experts.28.gate_and_up_proj.weight", "model.layers.25.mlp.experts.28.down_proj.weight", "model.layers.25.mlp.experts.29.gate_and_up_proj.weight", "model.layers.25.mlp.experts.29.down_proj.weight", "model.layers.25.mlp.experts.30.gate_and_up_proj.weight", "model.layers.25.mlp.experts.30.down_proj.weight", "model.layers.25.mlp.experts.31.gate_and_up_proj.weight", "model.layers.25.mlp.experts.31.down_proj.weight", "model.layers.25.mlp.experts.32.gate_and_up_proj.weight", "model.layers.25.mlp.experts.32.down_proj.weight", "model.layers.25.mlp.experts.33.gate_and_up_proj.weight", "model.layers.25.mlp.experts.33.down_proj.weight", "model.layers.25.mlp.experts.34.gate_and_up_proj.weight", "model.layers.25.mlp.experts.34.down_proj.weight", "model.layers.25.mlp.experts.35.gate_and_up_proj.weight", "model.layers.25.mlp.experts.35.down_proj.weight", "model.layers.25.mlp.experts.36.gate_and_up_proj.weight", "model.layers.25.mlp.experts.36.down_proj.weight", "model.layers.25.mlp.experts.37.gate_and_up_proj.weight", "model.layers.25.mlp.experts.37.down_proj.weight", "model.layers.25.mlp.experts.38.gate_and_up_proj.weight", "model.layers.25.mlp.experts.38.down_proj.weight", "model.layers.25.mlp.experts.39.gate_and_up_proj.weight", "model.layers.25.mlp.experts.39.down_proj.weight", "model.layers.25.mlp.experts.40.gate_and_up_proj.weight", "model.layers.25.mlp.experts.40.down_proj.weight", "model.layers.25.mlp.experts.41.gate_and_up_proj.weight", "model.layers.25.mlp.experts.41.down_proj.weight", "model.layers.25.mlp.experts.42.gate_and_up_proj.weight", "model.layers.25.mlp.experts.42.down_proj.weight", "model.layers.25.mlp.experts.43.gate_and_up_proj.weight", "model.layers.25.mlp.experts.43.down_proj.weight", "model.layers.25.mlp.experts.44.gate_and_up_proj.weight", "model.layers.25.mlp.experts.44.down_proj.weight", "model.layers.25.mlp.experts.45.gate_and_up_proj.weight", "model.layers.25.mlp.experts.45.down_proj.weight", "model.layers.25.mlp.experts.46.gate_and_up_proj.weight", "model.layers.25.mlp.experts.46.down_proj.weight", "model.layers.25.mlp.experts.47.gate_and_up_proj.weight", "model.layers.25.mlp.experts.47.down_proj.weight", "model.layers.25.mlp.experts.48.gate_and_up_proj.weight", "model.layers.25.mlp.experts.48.down_proj.weight", "model.layers.25.mlp.experts.49.gate_and_up_proj.weight", "model.layers.25.mlp.experts.49.down_proj.weight", "model.layers.25.mlp.experts.50.gate_and_up_proj.weight", "model.layers.25.mlp.experts.50.down_proj.weight", "model.layers.25.mlp.experts.51.gate_and_up_proj.weight", "model.layers.25.mlp.experts.51.down_proj.weight", "model.layers.25.mlp.experts.52.gate_and_up_proj.weight", "model.layers.25.mlp.experts.52.down_proj.weight", "model.layers.25.mlp.experts.53.gate_and_up_proj.weight", "model.layers.25.mlp.experts.53.down_proj.weight", "model.layers.25.mlp.experts.54.gate_and_up_proj.weight", "model.layers.25.mlp.experts.54.down_proj.weight", "model.layers.25.mlp.experts.55.gate_and_up_proj.weight", "model.layers.25.mlp.experts.55.down_proj.weight", "model.layers.25.mlp.experts.56.gate_and_up_proj.weight", "model.layers.25.mlp.experts.56.down_proj.weight", "model.layers.25.mlp.experts.57.gate_and_up_proj.weight", "model.layers.25.mlp.experts.57.down_proj.weight", "model.layers.25.mlp.experts.58.gate_and_up_proj.weight", "model.layers.25.mlp.experts.58.down_proj.weight", "model.layers.25.mlp.experts.59.gate_and_up_proj.weight", "model.layers.25.mlp.experts.59.down_proj.weight", "model.layers.25.mlp.experts.60.gate_and_up_proj.weight", "model.layers.25.mlp.experts.60.down_proj.weight", "model.layers.25.mlp.experts.61.gate_and_up_proj.weight", "model.layers.25.mlp.experts.61.down_proj.weight", "model.layers.25.mlp.experts.62.gate_and_up_proj.weight", "model.layers.25.mlp.experts.62.down_proj.weight", "model.layers.25.mlp.experts.63.gate_and_up_proj.weight", "model.layers.25.mlp.experts.63.down_proj.weight", "model.layers.26.self_attn.qkv_proj.weight", "model.layers.26.self_attn.o_proj.weight", "model.layers.26.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.26.mlp.shared_mlp.down_proj.weight", "model.layers.26.mlp.experts.0.gate_and_up_proj.weight", "model.layers.26.mlp.experts.0.down_proj.weight", "model.layers.26.mlp.experts.1.gate_and_up_proj.weight", "model.layers.26.mlp.experts.1.down_proj.weight", "model.layers.26.mlp.experts.2.gate_and_up_proj.weight", "model.layers.26.mlp.experts.2.down_proj.weight", "model.layers.26.mlp.experts.3.gate_and_up_proj.weight", "model.layers.26.mlp.experts.3.down_proj.weight", "model.layers.26.mlp.experts.4.gate_and_up_proj.weight", "model.layers.26.mlp.experts.4.down_proj.weight", "model.layers.26.mlp.experts.5.gate_and_up_proj.weight", "model.layers.26.mlp.experts.5.down_proj.weight", "model.layers.26.mlp.experts.6.gate_and_up_proj.weight", "model.layers.26.mlp.experts.6.down_proj.weight", "model.layers.26.mlp.experts.7.gate_and_up_proj.weight", "model.layers.26.mlp.experts.7.down_proj.weight", "model.layers.26.mlp.experts.8.gate_and_up_proj.weight", "model.layers.26.mlp.experts.8.down_proj.weight", "model.layers.26.mlp.experts.9.gate_and_up_proj.weight", "model.layers.26.mlp.experts.9.down_proj.weight", "model.layers.26.mlp.experts.10.gate_and_up_proj.weight", "model.layers.26.mlp.experts.10.down_proj.weight", "model.layers.26.mlp.experts.11.gate_and_up_proj.weight", "model.layers.26.mlp.experts.11.down_proj.weight", "model.layers.26.mlp.experts.12.gate_and_up_proj.weight", "model.layers.26.mlp.experts.12.down_proj.weight", "model.layers.26.mlp.experts.13.gate_and_up_proj.weight", "model.layers.26.mlp.experts.13.down_proj.weight", "model.layers.26.mlp.experts.14.gate_and_up_proj.weight", "model.layers.26.mlp.experts.14.down_proj.weight", "model.layers.26.mlp.experts.15.gate_and_up_proj.weight", "model.layers.26.mlp.experts.15.down_proj.weight", "model.layers.26.mlp.experts.16.gate_and_up_proj.weight", "model.layers.26.mlp.experts.16.down_proj.weight", "model.layers.26.mlp.experts.17.gate_and_up_proj.weight", "model.layers.26.mlp.experts.17.down_proj.weight", "model.layers.26.mlp.experts.18.gate_and_up_proj.weight", "model.layers.26.mlp.experts.18.down_proj.weight", "model.layers.26.mlp.experts.19.gate_and_up_proj.weight", "model.layers.26.mlp.experts.19.down_proj.weight", "model.layers.26.mlp.experts.20.gate_and_up_proj.weight", "model.layers.26.mlp.experts.20.down_proj.weight", "model.layers.26.mlp.experts.21.gate_and_up_proj.weight", "model.layers.26.mlp.experts.21.down_proj.weight", "model.layers.26.mlp.experts.22.gate_and_up_proj.weight", "model.layers.26.mlp.experts.22.down_proj.weight", "model.layers.26.mlp.experts.23.gate_and_up_proj.weight", "model.layers.26.mlp.experts.23.down_proj.weight", "model.layers.26.mlp.experts.24.gate_and_up_proj.weight", "model.layers.26.mlp.experts.24.down_proj.weight", "model.layers.26.mlp.experts.25.gate_and_up_proj.weight", "model.layers.26.mlp.experts.25.down_proj.weight", "model.layers.26.mlp.experts.26.gate_and_up_proj.weight", "model.layers.26.mlp.experts.26.down_proj.weight", "model.layers.26.mlp.experts.27.gate_and_up_proj.weight", "model.layers.26.mlp.experts.27.down_proj.weight", "model.layers.26.mlp.experts.28.gate_and_up_proj.weight", "model.layers.26.mlp.experts.28.down_proj.weight", "model.layers.26.mlp.experts.29.gate_and_up_proj.weight", "model.layers.26.mlp.experts.29.down_proj.weight", "model.layers.26.mlp.experts.30.gate_and_up_proj.weight", "model.layers.26.mlp.experts.30.down_proj.weight", "model.layers.26.mlp.experts.31.gate_and_up_proj.weight", "model.layers.26.mlp.experts.31.down_proj.weight", "model.layers.26.mlp.experts.32.gate_and_up_proj.weight", "model.layers.26.mlp.experts.32.down_proj.weight", "model.layers.26.mlp.experts.33.gate_and_up_proj.weight", "model.layers.26.mlp.experts.33.down_proj.weight", "model.layers.26.mlp.experts.34.gate_and_up_proj.weight", "model.layers.26.mlp.experts.34.down_proj.weight", "model.layers.26.mlp.experts.35.gate_and_up_proj.weight", "model.layers.26.mlp.experts.35.down_proj.weight", "model.layers.26.mlp.experts.36.gate_and_up_proj.weight", "model.layers.26.mlp.experts.36.down_proj.weight", "model.layers.26.mlp.experts.37.gate_and_up_proj.weight", "model.layers.26.mlp.experts.37.down_proj.weight", "model.layers.26.mlp.experts.38.gate_and_up_proj.weight", "model.layers.26.mlp.experts.38.down_proj.weight", "model.layers.26.mlp.experts.39.gate_and_up_proj.weight", "model.layers.26.mlp.experts.39.down_proj.weight", "model.layers.26.mlp.experts.40.gate_and_up_proj.weight", "model.layers.26.mlp.experts.40.down_proj.weight", "model.layers.26.mlp.experts.41.gate_and_up_proj.weight", "model.layers.26.mlp.experts.41.down_proj.weight", "model.layers.26.mlp.experts.42.gate_and_up_proj.weight", "model.layers.26.mlp.experts.42.down_proj.weight", "model.layers.26.mlp.experts.43.gate_and_up_proj.weight", "model.layers.26.mlp.experts.43.down_proj.weight", "model.layers.26.mlp.experts.44.gate_and_up_proj.weight", "model.layers.26.mlp.experts.44.down_proj.weight", "model.layers.26.mlp.experts.45.gate_and_up_proj.weight", "model.layers.26.mlp.experts.45.down_proj.weight", "model.layers.26.mlp.experts.46.gate_and_up_proj.weight", "model.layers.26.mlp.experts.46.down_proj.weight", "model.layers.26.mlp.experts.47.gate_and_up_proj.weight", "model.layers.26.mlp.experts.47.down_proj.weight", "model.layers.26.mlp.experts.48.gate_and_up_proj.weight", "model.layers.26.mlp.experts.48.down_proj.weight", "model.layers.26.mlp.experts.49.gate_and_up_proj.weight", "model.layers.26.mlp.experts.49.down_proj.weight", "model.layers.26.mlp.experts.50.gate_and_up_proj.weight", "model.layers.26.mlp.experts.50.down_proj.weight", "model.layers.26.mlp.experts.51.gate_and_up_proj.weight", "model.layers.26.mlp.experts.51.down_proj.weight", "model.layers.26.mlp.experts.52.gate_and_up_proj.weight", "model.layers.26.mlp.experts.52.down_proj.weight", "model.layers.26.mlp.experts.53.gate_and_up_proj.weight", "model.layers.26.mlp.experts.53.down_proj.weight", "model.layers.26.mlp.experts.54.gate_and_up_proj.weight", "model.layers.26.mlp.experts.54.down_proj.weight", "model.layers.26.mlp.experts.55.gate_and_up_proj.weight", "model.layers.26.mlp.experts.55.down_proj.weight", "model.layers.26.mlp.experts.56.gate_and_up_proj.weight", "model.layers.26.mlp.experts.56.down_proj.weight", "model.layers.26.mlp.experts.57.gate_and_up_proj.weight", "model.layers.26.mlp.experts.57.down_proj.weight", "model.layers.26.mlp.experts.58.gate_and_up_proj.weight", "model.layers.26.mlp.experts.58.down_proj.weight", "model.layers.26.mlp.experts.59.gate_and_up_proj.weight", "model.layers.26.mlp.experts.59.down_proj.weight", "model.layers.26.mlp.experts.60.gate_and_up_proj.weight", "model.layers.26.mlp.experts.60.down_proj.weight", "model.layers.26.mlp.experts.61.gate_and_up_proj.weight", "model.layers.26.mlp.experts.61.down_proj.weight", "model.layers.26.mlp.experts.62.gate_and_up_proj.weight", "model.layers.26.mlp.experts.62.down_proj.weight", "model.layers.26.mlp.experts.63.gate_and_up_proj.weight", "model.layers.26.mlp.experts.63.down_proj.weight", "model.layers.27.self_attn.qkv_proj.weight", "model.layers.27.self_attn.o_proj.weight", "model.layers.27.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.27.mlp.experts.0.gate_and_up_proj.weight", "model.layers.27.mlp.experts.0.down_proj.weight", "model.layers.27.mlp.experts.1.gate_and_up_proj.weight", "model.layers.27.mlp.experts.1.down_proj.weight", "model.layers.27.mlp.experts.2.gate_and_up_proj.weight", "model.layers.27.mlp.experts.2.down_proj.weight", "model.layers.27.mlp.experts.3.gate_and_up_proj.weight", "model.layers.27.mlp.experts.3.down_proj.weight", "model.layers.27.mlp.experts.4.gate_and_up_proj.weight", "model.layers.27.mlp.experts.4.down_proj.weight", "model.layers.27.mlp.experts.5.gate_and_up_proj.weight", "model.layers.27.mlp.experts.5.down_proj.weight", "model.layers.27.mlp.experts.6.gate_and_up_proj.weight", "model.layers.27.mlp.experts.6.down_proj.weight", "model.layers.27.mlp.experts.7.gate_and_up_proj.weight", "model.layers.27.mlp.experts.7.down_proj.weight", "model.layers.27.mlp.experts.8.gate_and_up_proj.weight", "model.layers.27.mlp.experts.8.down_proj.weight", "model.layers.27.mlp.experts.9.gate_and_up_proj.weight", "model.layers.27.mlp.experts.9.down_proj.weight", "model.layers.27.mlp.experts.10.gate_and_up_proj.weight", "model.layers.27.mlp.experts.10.down_proj.weight", "model.layers.27.mlp.experts.11.gate_and_up_proj.weight", "model.layers.27.mlp.experts.11.down_proj.weight", "model.layers.27.mlp.experts.12.gate_and_up_proj.weight", "model.layers.27.mlp.experts.12.down_proj.weight", "model.layers.27.mlp.experts.13.gate_and_up_proj.weight", "model.layers.27.mlp.experts.13.down_proj.weight", "model.layers.27.mlp.experts.14.gate_and_up_proj.weight", "model.layers.27.mlp.experts.14.down_proj.weight", "model.layers.27.mlp.experts.15.gate_and_up_proj.weight", "model.layers.27.mlp.experts.15.down_proj.weight", "model.layers.27.mlp.experts.16.gate_and_up_proj.weight", "model.layers.27.mlp.experts.16.down_proj.weight", "model.layers.27.mlp.experts.17.gate_and_up_proj.weight", "model.layers.27.mlp.experts.17.down_proj.weight", "model.layers.27.mlp.experts.18.gate_and_up_proj.weight", "model.layers.27.mlp.experts.18.down_proj.weight", "model.layers.27.mlp.experts.19.gate_and_up_proj.weight", "model.layers.27.mlp.experts.19.down_proj.weight", "model.layers.27.mlp.experts.20.gate_and_up_proj.weight", "model.layers.27.mlp.experts.20.down_proj.weight", "model.layers.27.mlp.experts.21.gate_and_up_proj.weight", "model.layers.27.mlp.experts.21.down_proj.weight", "model.layers.27.mlp.experts.22.gate_and_up_proj.weight", "model.layers.27.mlp.experts.22.down_proj.weight", "model.layers.27.mlp.experts.23.gate_and_up_proj.weight", "model.layers.27.mlp.experts.23.down_proj.weight", "model.layers.27.mlp.experts.24.gate_and_up_proj.weight", "model.layers.27.mlp.experts.24.down_proj.weight", "model.layers.27.mlp.experts.25.gate_and_up_proj.weight", "model.layers.27.mlp.experts.25.down_proj.weight", "model.layers.27.mlp.experts.26.gate_and_up_proj.weight", "model.layers.27.mlp.experts.26.down_proj.weight", "model.layers.27.mlp.experts.27.gate_and_up_proj.weight", "model.layers.27.mlp.experts.27.down_proj.weight", "model.layers.27.mlp.experts.28.gate_and_up_proj.weight", "model.layers.27.mlp.experts.28.down_proj.weight", "model.layers.27.mlp.experts.29.gate_and_up_proj.weight", "model.layers.27.mlp.experts.29.down_proj.weight", "model.layers.27.mlp.experts.30.gate_and_up_proj.weight", "model.layers.27.mlp.experts.30.down_proj.weight", "model.layers.27.mlp.experts.31.gate_and_up_proj.weight", "model.layers.27.mlp.experts.31.down_proj.weight", "model.layers.27.mlp.experts.32.gate_and_up_proj.weight", "model.layers.27.mlp.experts.32.down_proj.weight", "model.layers.27.mlp.experts.33.gate_and_up_proj.weight", "model.layers.27.mlp.experts.33.down_proj.weight", "model.layers.27.mlp.experts.34.gate_and_up_proj.weight", "model.layers.27.mlp.experts.34.down_proj.weight", "model.layers.27.mlp.experts.35.gate_and_up_proj.weight", "model.layers.27.mlp.experts.35.down_proj.weight", "model.layers.27.mlp.experts.36.gate_and_up_proj.weight", "model.layers.27.mlp.experts.36.down_proj.weight", "model.layers.27.mlp.experts.37.gate_and_up_proj.weight", "model.layers.27.mlp.experts.37.down_proj.weight", "model.layers.27.mlp.experts.38.gate_and_up_proj.weight", "model.layers.27.mlp.experts.38.down_proj.weight", "model.layers.27.mlp.experts.39.gate_and_up_proj.weight", "model.layers.27.mlp.experts.39.down_proj.weight", "model.layers.27.mlp.experts.40.gate_and_up_proj.weight", "model.layers.27.mlp.experts.40.down_proj.weight", "model.layers.27.mlp.experts.41.gate_and_up_proj.weight", "model.layers.27.mlp.experts.41.down_proj.weight", "model.layers.27.mlp.experts.42.gate_and_up_proj.weight", "model.layers.27.mlp.experts.42.down_proj.weight", "model.layers.27.mlp.experts.43.gate_and_up_proj.weight", "model.layers.27.mlp.experts.43.down_proj.weight", "model.layers.27.mlp.experts.44.gate_and_up_proj.weight", "model.layers.27.mlp.experts.44.down_proj.weight", "model.layers.27.mlp.experts.45.gate_and_up_proj.weight", "model.layers.27.mlp.experts.45.down_proj.weight", "model.layers.27.mlp.experts.46.gate_and_up_proj.weight", "model.layers.27.mlp.experts.46.down_proj.weight", "model.layers.27.mlp.experts.47.gate_and_up_proj.weight", "model.layers.27.mlp.experts.47.down_proj.weight", "model.layers.27.mlp.experts.48.gate_and_up_proj.weight", "model.layers.27.mlp.experts.48.down_proj.weight", "model.layers.27.mlp.experts.49.gate_and_up_proj.weight", "model.layers.27.mlp.experts.49.down_proj.weight", "model.layers.27.mlp.experts.50.gate_and_up_proj.weight", "model.layers.27.mlp.experts.50.down_proj.weight", "model.layers.27.mlp.experts.51.gate_and_up_proj.weight", "model.layers.27.mlp.experts.51.down_proj.weight", "model.layers.27.mlp.experts.52.gate_and_up_proj.weight", "model.layers.27.mlp.experts.52.down_proj.weight", "model.layers.27.mlp.experts.53.gate_and_up_proj.weight", "model.layers.27.mlp.experts.53.down_proj.weight", "model.layers.27.mlp.experts.54.gate_and_up_proj.weight", "model.layers.27.mlp.experts.54.down_proj.weight", "model.layers.27.mlp.experts.55.gate_and_up_proj.weight", "model.layers.27.mlp.experts.55.down_proj.weight", "model.layers.27.mlp.experts.56.gate_and_up_proj.weight", "model.layers.27.mlp.experts.56.down_proj.weight", "model.layers.27.mlp.experts.57.gate_and_up_proj.weight", "model.layers.27.mlp.experts.57.down_proj.weight", "model.layers.27.mlp.experts.58.gate_and_up_proj.weight", "model.layers.27.mlp.experts.58.down_proj.weight", "model.layers.27.mlp.experts.59.gate_and_up_proj.weight", "model.layers.27.mlp.experts.59.down_proj.weight", "model.layers.27.mlp.experts.60.gate_and_up_proj.weight", "model.layers.27.mlp.experts.60.down_proj.weight", "model.layers.27.mlp.experts.61.gate_and_up_proj.weight", "model.layers.27.mlp.experts.61.down_proj.weight", "model.layers.27.mlp.experts.62.gate_and_up_proj.weight", "model.layers.27.mlp.experts.62.down_proj.weight", "model.layers.27.mlp.experts.63.gate_and_up_proj.weight", "model.layers.27.mlp.experts.63.down_proj.weight", "model.layers.28.self_attn.qkv_proj.weight", "model.layers.28.self_attn.o_proj.weight", "model.layers.28.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.28.mlp.experts.0.gate_and_up_proj.weight", "model.layers.28.mlp.experts.0.down_proj.weight", "model.layers.28.mlp.experts.1.gate_and_up_proj.weight", "model.layers.28.mlp.experts.1.down_proj.weight", "model.layers.28.mlp.experts.2.gate_and_up_proj.weight", "model.layers.28.mlp.experts.2.down_proj.weight", "model.layers.28.mlp.experts.3.gate_and_up_proj.weight", "model.layers.28.mlp.experts.3.down_proj.weight", "model.layers.28.mlp.experts.4.gate_and_up_proj.weight", "model.layers.28.mlp.experts.4.down_proj.weight", "model.layers.28.mlp.experts.5.gate_and_up_proj.weight", "model.layers.28.mlp.experts.5.down_proj.weight", "model.layers.28.mlp.experts.6.gate_and_up_proj.weight", "model.layers.28.mlp.experts.6.down_proj.weight", "model.layers.28.mlp.experts.7.gate_and_up_proj.weight", "model.layers.28.mlp.experts.7.down_proj.weight", "model.layers.28.mlp.experts.8.gate_and_up_proj.weight", "model.layers.28.mlp.experts.8.down_proj.weight", "model.layers.28.mlp.experts.9.gate_and_up_proj.weight", "model.layers.28.mlp.experts.9.down_proj.weight", "model.layers.28.mlp.experts.10.gate_and_up_proj.weight", "model.layers.28.mlp.experts.10.down_proj.weight", "model.layers.28.mlp.experts.11.gate_and_up_proj.weight", "model.layers.28.mlp.experts.11.down_proj.weight", "model.layers.28.mlp.experts.12.gate_and_up_proj.weight", "model.layers.28.mlp.experts.12.down_proj.weight", "model.layers.28.mlp.experts.13.gate_and_up_proj.weight", "model.layers.28.mlp.experts.13.down_proj.weight", "model.layers.28.mlp.experts.14.gate_and_up_proj.weight", "model.layers.28.mlp.experts.14.down_proj.weight", "model.layers.28.mlp.experts.15.gate_and_up_proj.weight", "model.layers.28.mlp.experts.15.down_proj.weight", "model.layers.28.mlp.experts.16.gate_and_up_proj.weight", "model.layers.28.mlp.experts.16.down_proj.weight", "model.layers.28.mlp.experts.17.gate_and_up_proj.weight", "model.layers.28.mlp.experts.17.down_proj.weight", "model.layers.28.mlp.experts.18.gate_and_up_proj.weight", "model.layers.28.mlp.experts.18.down_proj.weight", "model.layers.28.mlp.experts.19.gate_and_up_proj.weight", "model.layers.28.mlp.experts.19.down_proj.weight", "model.layers.28.mlp.experts.20.gate_and_up_proj.weight", "model.layers.28.mlp.experts.20.down_proj.weight", "model.layers.28.mlp.experts.21.gate_and_up_proj.weight", "model.layers.28.mlp.experts.21.down_proj.weight", "model.layers.28.mlp.experts.22.gate_and_up_proj.weight", "model.layers.28.mlp.experts.22.down_proj.weight", "model.layers.28.mlp.experts.23.gate_and_up_proj.weight", "model.layers.28.mlp.experts.23.down_proj.weight", "model.layers.28.mlp.experts.24.gate_and_up_proj.weight", "model.layers.28.mlp.experts.24.down_proj.weight", "model.layers.28.mlp.experts.25.gate_and_up_proj.weight", "model.layers.28.mlp.experts.25.down_proj.weight", "model.layers.28.mlp.experts.26.gate_and_up_proj.weight", "model.layers.28.mlp.experts.26.down_proj.weight", "model.layers.28.mlp.experts.27.gate_and_up_proj.weight", "model.layers.28.mlp.experts.27.down_proj.weight", "model.layers.28.mlp.experts.28.gate_and_up_proj.weight", "model.layers.28.mlp.experts.28.down_proj.weight", "model.layers.28.mlp.experts.29.gate_and_up_proj.weight", "model.layers.28.mlp.experts.29.down_proj.weight", "model.layers.28.mlp.experts.30.gate_and_up_proj.weight", "model.layers.28.mlp.experts.30.down_proj.weight", "model.layers.28.mlp.experts.31.gate_and_up_proj.weight", "model.layers.28.mlp.experts.31.down_proj.weight", "model.layers.28.mlp.experts.32.gate_and_up_proj.weight", "model.layers.28.mlp.experts.32.down_proj.weight", "model.layers.28.mlp.experts.33.gate_and_up_proj.weight", "model.layers.28.mlp.experts.33.down_proj.weight", "model.layers.28.mlp.experts.34.gate_and_up_proj.weight", "model.layers.28.mlp.experts.34.down_proj.weight", "model.layers.28.mlp.experts.35.gate_and_up_proj.weight", "model.layers.28.mlp.experts.35.down_proj.weight", "model.layers.28.mlp.experts.36.gate_and_up_proj.weight", "model.layers.28.mlp.experts.36.down_proj.weight", "model.layers.28.mlp.experts.37.gate_and_up_proj.weight", "model.layers.28.mlp.experts.37.down_proj.weight", "model.layers.28.mlp.experts.38.gate_and_up_proj.weight", "model.layers.28.mlp.experts.38.down_proj.weight", "model.layers.28.mlp.experts.39.gate_and_up_proj.weight", "model.layers.28.mlp.experts.39.down_proj.weight", "model.layers.28.mlp.experts.40.gate_and_up_proj.weight", "model.layers.28.mlp.experts.40.down_proj.weight", "model.layers.28.mlp.experts.41.gate_and_up_proj.weight", "model.layers.28.mlp.experts.41.down_proj.weight", "model.layers.28.mlp.experts.42.gate_and_up_proj.weight", "model.layers.28.mlp.experts.42.down_proj.weight", "model.layers.28.mlp.experts.43.gate_and_up_proj.weight", "model.layers.28.mlp.experts.43.down_proj.weight", "model.layers.28.mlp.experts.44.gate_and_up_proj.weight", "model.layers.28.mlp.experts.44.down_proj.weight", "model.layers.28.mlp.experts.45.gate_and_up_proj.weight", "model.layers.28.mlp.experts.45.down_proj.weight", "model.layers.28.mlp.experts.46.gate_and_up_proj.weight", "model.layers.28.mlp.experts.46.down_proj.weight", "model.layers.28.mlp.experts.47.gate_and_up_proj.weight", "model.layers.28.mlp.experts.47.down_proj.weight", "model.layers.28.mlp.experts.48.gate_and_up_proj.weight", "model.layers.28.mlp.experts.48.down_proj.weight", "model.layers.28.mlp.experts.49.gate_and_up_proj.weight", "model.layers.28.mlp.experts.49.down_proj.weight", "model.layers.28.mlp.experts.50.gate_and_up_proj.weight", "model.layers.28.mlp.experts.50.down_proj.weight", "model.layers.28.mlp.experts.51.gate_and_up_proj.weight", "model.layers.28.mlp.experts.51.down_proj.weight", "model.layers.28.mlp.experts.52.gate_and_up_proj.weight", "model.layers.28.mlp.experts.52.down_proj.weight", "model.layers.28.mlp.experts.53.gate_and_up_proj.weight", "model.layers.28.mlp.experts.53.down_proj.weight", "model.layers.28.mlp.experts.54.gate_and_up_proj.weight", "model.layers.28.mlp.experts.54.down_proj.weight", "model.layers.28.mlp.experts.55.gate_and_up_proj.weight", "model.layers.28.mlp.experts.55.down_proj.weight", "model.layers.28.mlp.experts.56.gate_and_up_proj.weight", "model.layers.28.mlp.experts.56.down_proj.weight", "model.layers.28.mlp.experts.57.gate_and_up_proj.weight", "model.layers.28.mlp.experts.57.down_proj.weight", "model.layers.28.mlp.experts.58.gate_and_up_proj.weight", "model.layers.28.mlp.experts.58.down_proj.weight", "model.layers.28.mlp.experts.59.gate_and_up_proj.weight", "model.layers.28.mlp.experts.59.down_proj.weight", "model.layers.28.mlp.experts.60.gate_and_up_proj.weight", "model.layers.28.mlp.experts.60.down_proj.weight", "model.layers.28.mlp.experts.61.gate_and_up_proj.weight", "model.layers.28.mlp.experts.61.down_proj.weight", "model.layers.28.mlp.experts.62.gate_and_up_proj.weight", "model.layers.28.mlp.experts.62.down_proj.weight", "model.layers.28.mlp.experts.63.gate_and_up_proj.weight", "model.layers.28.mlp.experts.63.down_proj.weight", "model.layers.29.self_attn.qkv_proj.weight", "model.layers.29.self_attn.o_proj.weight", "model.layers.29.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.29.mlp.experts.0.gate_and_up_proj.weight", "model.layers.29.mlp.experts.0.down_proj.weight", "model.layers.29.mlp.experts.1.gate_and_up_proj.weight", "model.layers.29.mlp.experts.1.down_proj.weight", "model.layers.29.mlp.experts.2.gate_and_up_proj.weight", "model.layers.29.mlp.experts.2.down_proj.weight", "model.layers.29.mlp.experts.3.gate_and_up_proj.weight", "model.layers.29.mlp.experts.3.down_proj.weight", "model.layers.29.mlp.experts.4.gate_and_up_proj.weight", "model.layers.29.mlp.experts.4.down_proj.weight", "model.layers.29.mlp.experts.5.gate_and_up_proj.weight", "model.layers.29.mlp.experts.5.down_proj.weight", "model.layers.29.mlp.experts.6.gate_and_up_proj.weight", "model.layers.29.mlp.experts.6.down_proj.weight", "model.layers.29.mlp.experts.7.gate_and_up_proj.weight", "model.layers.29.mlp.experts.7.down_proj.weight", "model.layers.29.mlp.experts.8.gate_and_up_proj.weight", "model.layers.29.mlp.experts.8.down_proj.weight", "model.layers.29.mlp.experts.9.gate_and_up_proj.weight", "model.layers.29.mlp.experts.9.down_proj.weight", "model.layers.29.mlp.experts.10.gate_and_up_proj.weight", "model.layers.29.mlp.experts.10.down_proj.weight", "model.layers.29.mlp.experts.11.gate_and_up_proj.weight", "model.layers.29.mlp.experts.11.down_proj.weight", "model.layers.29.mlp.experts.12.gate_and_up_proj.weight", "model.layers.29.mlp.experts.12.down_proj.weight", "model.layers.29.mlp.experts.13.gate_and_up_proj.weight", "model.layers.29.mlp.experts.13.down_proj.weight", "model.layers.29.mlp.experts.14.gate_and_up_proj.weight", "model.layers.29.mlp.experts.14.down_proj.weight", "model.layers.29.mlp.experts.15.gate_and_up_proj.weight", "model.layers.29.mlp.experts.15.down_proj.weight", "model.layers.29.mlp.experts.16.gate_and_up_proj.weight", "model.layers.29.mlp.experts.16.down_proj.weight", "model.layers.29.mlp.experts.17.gate_and_up_proj.weight", "model.layers.29.mlp.experts.17.down_proj.weight", "model.layers.29.mlp.experts.18.gate_and_up_proj.weight", "model.layers.29.mlp.experts.18.down_proj.weight", "model.layers.29.mlp.experts.19.gate_and_up_proj.weight", "model.layers.29.mlp.experts.19.down_proj.weight", "model.layers.29.mlp.experts.20.gate_and_up_proj.weight", "model.layers.29.mlp.experts.20.down_proj.weight", "model.layers.29.mlp.experts.21.gate_and_up_proj.weight", "model.layers.29.mlp.experts.21.down_proj.weight", "model.layers.29.mlp.experts.22.gate_and_up_proj.weight", "model.layers.29.mlp.experts.22.down_proj.weight", "model.layers.29.mlp.experts.23.gate_and_up_proj.weight", "model.layers.29.mlp.experts.23.down_proj.weight", "model.layers.29.mlp.experts.24.gate_and_up_proj.weight", "model.layers.29.mlp.experts.24.down_proj.weight", "model.layers.29.mlp.experts.25.gate_and_up_proj.weight", "model.layers.29.mlp.experts.25.down_proj.weight", "model.layers.29.mlp.experts.26.gate_and_up_proj.weight", "model.layers.29.mlp.experts.26.down_proj.weight", "model.layers.29.mlp.experts.27.gate_and_up_proj.weight", "model.layers.29.mlp.experts.27.down_proj.weight", "model.layers.29.mlp.experts.28.gate_and_up_proj.weight", "model.layers.29.mlp.experts.28.down_proj.weight", "model.layers.29.mlp.experts.29.gate_and_up_proj.weight", "model.layers.29.mlp.experts.29.down_proj.weight", "model.layers.29.mlp.experts.30.gate_and_up_proj.weight", "model.layers.29.mlp.experts.30.down_proj.weight", "model.layers.29.mlp.experts.31.gate_and_up_proj.weight", "model.layers.29.mlp.experts.31.down_proj.weight", "model.layers.29.mlp.experts.32.gate_and_up_proj.weight", "model.layers.29.mlp.experts.32.down_proj.weight", "model.layers.29.mlp.experts.33.gate_and_up_proj.weight", "model.layers.29.mlp.experts.33.down_proj.weight", "model.layers.29.mlp.experts.34.gate_and_up_proj.weight", "model.layers.29.mlp.experts.34.down_proj.weight", "model.layers.29.mlp.experts.35.gate_and_up_proj.weight", "model.layers.29.mlp.experts.35.down_proj.weight", "model.layers.29.mlp.experts.36.gate_and_up_proj.weight", "model.layers.29.mlp.experts.36.down_proj.weight", "model.layers.29.mlp.experts.37.gate_and_up_proj.weight", "model.layers.29.mlp.experts.37.down_proj.weight", "model.layers.29.mlp.experts.38.gate_and_up_proj.weight", "model.layers.29.mlp.experts.38.down_proj.weight", "model.layers.29.mlp.experts.39.gate_and_up_proj.weight", "model.layers.29.mlp.experts.39.down_proj.weight", "model.layers.29.mlp.experts.40.gate_and_up_proj.weight", "model.layers.29.mlp.experts.40.down_proj.weight", "model.layers.29.mlp.experts.41.gate_and_up_proj.weight", "model.layers.29.mlp.experts.41.down_proj.weight", "model.layers.29.mlp.experts.42.gate_and_up_proj.weight", "model.layers.29.mlp.experts.42.down_proj.weight", "model.layers.29.mlp.experts.43.gate_and_up_proj.weight", "model.layers.29.mlp.experts.43.down_proj.weight", "model.layers.29.mlp.experts.44.gate_and_up_proj.weight", "model.layers.29.mlp.experts.44.down_proj.weight", "model.layers.29.mlp.experts.45.gate_and_up_proj.weight", "model.layers.29.mlp.experts.45.down_proj.weight", "model.layers.29.mlp.experts.46.gate_and_up_proj.weight", "model.layers.29.mlp.experts.46.down_proj.weight", "model.layers.29.mlp.experts.47.gate_and_up_proj.weight", "model.layers.29.mlp.experts.47.down_proj.weight", "model.layers.29.mlp.experts.48.gate_and_up_proj.weight", "model.layers.29.mlp.experts.48.down_proj.weight", "model.layers.29.mlp.experts.49.gate_and_up_proj.weight", "model.layers.29.mlp.experts.49.down_proj.weight", "model.layers.29.mlp.experts.50.gate_and_up_proj.weight", "model.layers.29.mlp.experts.50.down_proj.weight", "model.layers.29.mlp.experts.51.gate_and_up_proj.weight", "model.layers.29.mlp.experts.51.down_proj.weight", "model.layers.29.mlp.experts.52.gate_and_up_proj.weight", "model.layers.29.mlp.experts.52.down_proj.weight", "model.layers.29.mlp.experts.53.gate_and_up_proj.weight", "model.layers.29.mlp.experts.53.down_proj.weight", "model.layers.29.mlp.experts.54.gate_and_up_proj.weight", "model.layers.29.mlp.experts.54.down_proj.weight", "model.layers.29.mlp.experts.55.gate_and_up_proj.weight", "model.layers.29.mlp.experts.55.down_proj.weight", "model.layers.29.mlp.experts.56.gate_and_up_proj.weight", "model.layers.29.mlp.experts.56.down_proj.weight", "model.layers.29.mlp.experts.57.gate_and_up_proj.weight", "model.layers.29.mlp.experts.57.down_proj.weight", "model.layers.29.mlp.experts.58.gate_and_up_proj.weight", "model.layers.29.mlp.experts.58.down_proj.weight", "model.layers.29.mlp.experts.59.gate_and_up_proj.weight", "model.layers.29.mlp.experts.59.down_proj.weight", "model.layers.29.mlp.experts.60.gate_and_up_proj.weight", "model.layers.29.mlp.experts.60.down_proj.weight", "model.layers.29.mlp.experts.61.gate_and_up_proj.weight", "model.layers.29.mlp.experts.61.down_proj.weight", "model.layers.29.mlp.experts.62.gate_and_up_proj.weight", "model.layers.29.mlp.experts.62.down_proj.weight", "model.layers.29.mlp.experts.63.gate_and_up_proj.weight", "model.layers.29.mlp.experts.63.down_proj.weight", "model.layers.30.self_attn.qkv_proj.weight", "model.layers.30.self_attn.o_proj.weight", "model.layers.30.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.30.mlp.experts.0.gate_and_up_proj.weight", "model.layers.30.mlp.experts.0.down_proj.weight", "model.layers.30.mlp.experts.1.gate_and_up_proj.weight", "model.layers.30.mlp.experts.1.down_proj.weight", "model.layers.30.mlp.experts.2.gate_and_up_proj.weight", "model.layers.30.mlp.experts.2.down_proj.weight", "model.layers.30.mlp.experts.3.gate_and_up_proj.weight", "model.layers.30.mlp.experts.3.down_proj.weight", "model.layers.30.mlp.experts.4.gate_and_up_proj.weight", "model.layers.30.mlp.experts.4.down_proj.weight", "model.layers.30.mlp.experts.5.gate_and_up_proj.weight", "model.layers.30.mlp.experts.5.down_proj.weight", "model.layers.30.mlp.experts.6.gate_and_up_proj.weight", "model.layers.30.mlp.experts.6.down_proj.weight", "model.layers.30.mlp.experts.7.gate_and_up_proj.weight", "model.layers.30.mlp.experts.7.down_proj.weight", "model.layers.30.mlp.experts.8.gate_and_up_proj.weight", "model.layers.30.mlp.experts.8.down_proj.weight", "model.layers.30.mlp.experts.9.gate_and_up_proj.weight", "model.layers.30.mlp.experts.9.down_proj.weight", "model.layers.30.mlp.experts.10.gate_and_up_proj.weight", "model.layers.30.mlp.experts.10.down_proj.weight", "model.layers.30.mlp.experts.11.gate_and_up_proj.weight", "model.layers.30.mlp.experts.11.down_proj.weight", "model.layers.30.mlp.experts.12.gate_and_up_proj.weight", "model.layers.30.mlp.experts.12.down_proj.weight", "model.layers.30.mlp.experts.13.gate_and_up_proj.weight", "model.layers.30.mlp.experts.13.down_proj.weight", "model.layers.30.mlp.experts.14.gate_and_up_proj.weight", "model.layers.30.mlp.experts.14.down_proj.weight", "model.layers.30.mlp.experts.15.gate_and_up_proj.weight", "model.layers.30.mlp.experts.15.down_proj.weight", "model.layers.30.mlp.experts.16.gate_and_up_proj.weight", "model.layers.30.mlp.experts.16.down_proj.weight", "model.layers.30.mlp.experts.17.gate_and_up_proj.weight", "model.layers.30.mlp.experts.17.down_proj.weight", "model.layers.30.mlp.experts.18.gate_and_up_proj.weight", "model.layers.30.mlp.experts.18.down_proj.weight", "model.layers.30.mlp.experts.19.gate_and_up_proj.weight", "model.layers.30.mlp.experts.19.down_proj.weight", "model.layers.30.mlp.experts.20.gate_and_up_proj.weight", "model.layers.30.mlp.experts.20.down_proj.weight", "model.layers.30.mlp.experts.21.gate_and_up_proj.weight", "model.layers.30.mlp.experts.21.down_proj.weight", "model.layers.30.mlp.experts.22.gate_and_up_proj.weight", "model.layers.30.mlp.experts.22.down_proj.weight", "model.layers.30.mlp.experts.23.gate_and_up_proj.weight", "model.layers.30.mlp.experts.23.down_proj.weight", "model.layers.30.mlp.experts.24.gate_and_up_proj.weight", "model.layers.30.mlp.experts.24.down_proj.weight", "model.layers.30.mlp.experts.25.gate_and_up_proj.weight", "model.layers.30.mlp.experts.25.down_proj.weight", "model.layers.30.mlp.experts.26.gate_and_up_proj.weight", "model.layers.30.mlp.experts.26.down_proj.weight", "model.layers.30.mlp.experts.27.gate_and_up_proj.weight", "model.layers.30.mlp.experts.27.down_proj.weight", "model.layers.30.mlp.experts.28.gate_and_up_proj.weight", "model.layers.30.mlp.experts.28.down_proj.weight", "model.layers.30.mlp.experts.29.gate_and_up_proj.weight", "model.layers.30.mlp.experts.29.down_proj.weight", "model.layers.30.mlp.experts.30.gate_and_up_proj.weight", "model.layers.30.mlp.experts.30.down_proj.weight", "model.layers.30.mlp.experts.31.gate_and_up_proj.weight", "model.layers.30.mlp.experts.31.down_proj.weight", "model.layers.30.mlp.experts.32.gate_and_up_proj.weight", "model.layers.30.mlp.experts.32.down_proj.weight", "model.layers.30.mlp.experts.33.gate_and_up_proj.weight", "model.layers.30.mlp.experts.33.down_proj.weight", "model.layers.30.mlp.experts.34.gate_and_up_proj.weight", "model.layers.30.mlp.experts.34.down_proj.weight", "model.layers.30.mlp.experts.35.gate_and_up_proj.weight", "model.layers.30.mlp.experts.35.down_proj.weight", "model.layers.30.mlp.experts.36.gate_and_up_proj.weight", "model.layers.30.mlp.experts.36.down_proj.weight", "model.layers.30.mlp.experts.37.gate_and_up_proj.weight", "model.layers.30.mlp.experts.37.down_proj.weight", "model.layers.30.mlp.experts.38.gate_and_up_proj.weight", "model.layers.30.mlp.experts.38.down_proj.weight", "model.layers.30.mlp.experts.39.gate_and_up_proj.weight", "model.layers.30.mlp.experts.39.down_proj.weight", "model.layers.30.mlp.experts.40.gate_and_up_proj.weight", "model.layers.30.mlp.experts.40.down_proj.weight", "model.layers.30.mlp.experts.41.gate_and_up_proj.weight", "model.layers.30.mlp.experts.41.down_proj.weight", "model.layers.30.mlp.experts.42.gate_and_up_proj.weight", "model.layers.30.mlp.experts.42.down_proj.weight", "model.layers.30.mlp.experts.43.gate_and_up_proj.weight", "model.layers.30.mlp.experts.43.down_proj.weight", "model.layers.30.mlp.experts.44.gate_and_up_proj.weight", "model.layers.30.mlp.experts.44.down_proj.weight", "model.layers.30.mlp.experts.45.gate_and_up_proj.weight", "model.layers.30.mlp.experts.45.down_proj.weight", "model.layers.30.mlp.experts.46.gate_and_up_proj.weight", "model.layers.30.mlp.experts.46.down_proj.weight", "model.layers.30.mlp.experts.47.gate_and_up_proj.weight", "model.layers.30.mlp.experts.47.down_proj.weight", "model.layers.30.mlp.experts.48.gate_and_up_proj.weight", "model.layers.30.mlp.experts.48.down_proj.weight", "model.layers.30.mlp.experts.49.gate_and_up_proj.weight", "model.layers.30.mlp.experts.49.down_proj.weight", "model.layers.30.mlp.experts.50.gate_and_up_proj.weight", "model.layers.30.mlp.experts.50.down_proj.weight", "model.layers.30.mlp.experts.51.gate_and_up_proj.weight", "model.layers.30.mlp.experts.51.down_proj.weight", "model.layers.30.mlp.experts.52.gate_and_up_proj.weight", "model.layers.30.mlp.experts.52.down_proj.weight", "model.layers.30.mlp.experts.53.gate_and_up_proj.weight", "model.layers.30.mlp.experts.53.down_proj.weight", "model.layers.30.mlp.experts.54.gate_and_up_proj.weight", "model.layers.30.mlp.experts.54.down_proj.weight", "model.layers.30.mlp.experts.55.gate_and_up_proj.weight", "model.layers.30.mlp.experts.55.down_proj.weight", "model.layers.30.mlp.experts.56.gate_and_up_proj.weight", "model.layers.30.mlp.experts.56.down_proj.weight", "model.layers.30.mlp.experts.57.gate_and_up_proj.weight", "model.layers.30.mlp.experts.57.down_proj.weight", "model.layers.30.mlp.experts.58.gate_and_up_proj.weight", "model.layers.30.mlp.experts.58.down_proj.weight", "model.layers.30.mlp.experts.59.gate_and_up_proj.weight", "model.layers.30.mlp.experts.59.down_proj.weight", "model.layers.30.mlp.experts.60.gate_and_up_proj.weight", "model.layers.30.mlp.experts.60.down_proj.weight", "model.layers.30.mlp.experts.61.gate_and_up_proj.weight", "model.layers.30.mlp.experts.61.down_proj.weight", "model.layers.30.mlp.experts.62.gate_and_up_proj.weight", "model.layers.30.mlp.experts.62.down_proj.weight", "model.layers.30.mlp.experts.63.gate_and_up_proj.weight", "model.layers.30.mlp.experts.63.down_proj.weight", "model.layers.31.self_attn.qkv_proj.weight", "model.layers.31.self_attn.o_proj.weight", "model.layers.31.mlp.shared_mlp.gate_and_up_proj.weight", "model.layers.31.mlp.gate.wg.weight", "model.layers.31.mlp.experts.0.gate_and_up_proj.weight", "model.layers.31.mlp.experts.0.down_proj.weight", "model.layers.31.mlp.experts.1.gate_and_up_proj.weight", "model.layers.31.mlp.experts.1.down_proj.weight", "model.layers.31.mlp.experts.2.gate_and_up_proj.weight", "model.layers.31.mlp.experts.2.down_proj.weight", "model.layers.31.mlp.experts.3.gate_and_up_proj.weight", "model.layers.31.mlp.experts.3.down_proj.weight", "model.layers.31.mlp.experts.4.gate_and_up_proj.weight", "model.layers.31.mlp.experts.4.down_proj.weight", "model.layers.31.mlp.experts.5.gate_and_up_proj.weight", "model.layers.31.mlp.experts.5.down_proj.weight", "model.layers.31.mlp.experts.6.gate_and_up_proj.weight", "model.layers.31.mlp.experts.6.down_proj.weight", "model.layers.31.mlp.experts.7.gate_and_up_proj.weight", "model.layers.31.mlp.experts.7.down_proj.weight", "model.layers.31.mlp.experts.8.gate_and_up_proj.weight", "model.layers.31.mlp.experts.8.down_proj.weight", "model.layers.31.mlp.experts.9.gate_and_up_proj.weight", "model.layers.31.mlp.experts.9.down_proj.weight", "model.layers.31.mlp.experts.10.gate_and_up_proj.weight", "model.layers.31.mlp.experts.10.down_proj.weight", "model.layers.31.mlp.experts.11.gate_and_up_proj.weight", "model.layers.31.mlp.experts.11.down_proj.weight", "model.layers.31.mlp.experts.12.gate_and_up_proj.weight", "model.layers.31.mlp.experts.12.down_proj.weight", "model.layers.31.mlp.experts.13.gate_and_up_proj.weight", "model.layers.31.mlp.experts.13.down_proj.weight", "model.layers.31.mlp.experts.14.gate_and_up_proj.weight", "model.layers.31.mlp.experts.14.down_proj.weight", "model.layers.31.mlp.experts.15.gate_and_up_proj.weight", "model.layers.31.mlp.experts.15.down_proj.weight", "model.layers.31.mlp.experts.16.gate_and_up_proj.weight", "model.layers.31.mlp.experts.16.down_proj.weight", "model.layers.31.mlp.experts.17.gate_and_up_proj.weight", "model.layers.31.mlp.experts.17.down_proj.weight", "model.layers.31.mlp.experts.18.gate_and_up_proj.weight", "model.layers.31.mlp.experts.18.down_proj.weight", "model.layers.31.mlp.experts.19.gate_and_up_proj.weight", "model.layers.31.mlp.experts.19.down_proj.weight", "model.layers.31.mlp.experts.20.gate_and_up_proj.weight", "model.layers.31.mlp.experts.20.down_proj.weight", "model.layers.31.mlp.experts.21.gate_and_up_proj.weight", "model.layers.31.mlp.experts.21.down_proj.weight", "model.layers.31.mlp.experts.22.gate_and_up_proj.weight", "model.layers.31.mlp.experts.22.down_proj.weight", "model.layers.31.mlp.experts.23.gate_and_up_proj.weight", "model.layers.31.mlp.experts.23.down_proj.weight", "model.layers.31.mlp.experts.24.gate_and_up_proj.weight", "model.layers.31.mlp.experts.24.down_proj.weight", "model.layers.31.mlp.experts.25.gate_and_up_proj.weight", "model.layers.31.mlp.experts.25.down_proj.weight", "model.layers.31.mlp.experts.26.gate_and_up_proj.weight", "model.layers.31.mlp.experts.26.down_proj.weight", "model.layers.31.mlp.experts.27.gate_and_up_proj.weight", "model.layers.31.mlp.experts.27.down_proj.weight", "model.layers.31.mlp.experts.28.gate_and_up_proj.weight", "model.layers.31.mlp.experts.28.down_proj.weight", "model.layers.31.mlp.experts.29.gate_and_up_proj.weight", "model.layers.31.mlp.experts.29.down_proj.weight", "model.layers.31.mlp.experts.30.gate_and_up_proj.weight", "model.layers.31.mlp.experts.30.down_proj.weight", "model.layers.31.mlp.experts.31.gate_and_up_proj.weight", "model.layers.31.mlp.experts.31.down_proj.weight", "model.layers.31.mlp.experts.32.gate_and_up_proj.weight", "model.layers.31.mlp.experts.32.down_proj.weight", "model.layers.31.mlp.experts.33.gate_and_up_proj.weight", "model.layers.31.mlp.experts.33.down_proj.weight", "model.layers.31.mlp.experts.34.gate_and_up_proj.weight", "model.layers.31.mlp.experts.34.down_proj.weight", "model.layers.31.mlp.experts.35.gate_and_up_proj.weight", "model.layers.31.mlp.experts.35.down_proj.weight", "model.layers.31.mlp.experts.36.gate_and_up_proj.weight", "model.layers.31.mlp.experts.36.down_proj.weight", "model.layers.31.mlp.experts.37.gate_and_up_proj.weight", "model.layers.31.mlp.experts.37.down_proj.weight", "model.layers.31.mlp.experts.38.gate_and_up_proj.weight", "model.layers.31.mlp.experts.38.down_proj.weight", "model.layers.31.mlp.experts.39.gate_and_up_proj.weight", "model.layers.31.mlp.experts.39.down_proj.weight", "model.layers.31.mlp.experts.40.gate_and_up_proj.weight", "model.layers.31.mlp.experts.40.down_proj.weight", "model.layers.31.mlp.experts.41.gate_and_up_proj.weight", "model.layers.31.mlp.experts.41.down_proj.weight", "model.layers.31.mlp.experts.42.gate_and_up_proj.weight", "model.layers.31.mlp.experts.42.down_proj.weight", "model.layers.31.mlp.experts.43.gate_and_up_proj.weight", "model.layers.31.mlp.experts.43.down_proj.weight", "model.layers.31.mlp.experts.44.gate_and_up_proj.weight", "model.layers.31.mlp.experts.44.down_proj.weight", "model.layers.31.mlp.experts.45.gate_and_up_proj.weight", "model.layers.31.mlp.experts.45.down_proj.weight", "model.layers.31.mlp.experts.46.gate_and_up_proj.weight", "model.layers.31.mlp.experts.46.down_proj.weight", "model.layers.31.mlp.experts.47.gate_and_up_proj.weight", "model.layers.31.mlp.experts.47.down_proj.weight", "model.layers.31.mlp.experts.48.gate_and_up_proj.weight", "model.layers.31.mlp.experts.48.down_proj.weight", "model.layers.31.mlp.experts.49.gate_and_up_proj.weight", "model.layers.31.mlp.experts.49.down_proj.weight", "model.layers.31.mlp.experts.50.gate_and_up_proj.weight", "model.layers.31.mlp.experts.50.down_proj.weight", "model.layers.31.mlp.experts.51.gate_and_up_proj.weight", "model.layers.31.mlp.experts.51.down_proj.weight", "model.layers.31.mlp.experts.52.gate_and_up_proj.weight", "model.layers.31.mlp.experts.52.down_proj.weight", "model.layers.31.mlp.experts.53.gate_and_up_proj.weight", "model.layers.31.mlp.experts.53.down_proj.weight", "model.layers.31.mlp.experts.54.gate_and_up_proj.weight", "model.layers.31.mlp.experts.54.down_proj.weight", "model.layers.31.mlp.experts.55.gate_and_up_proj.weight", "model.layers.31.mlp.experts.55.down_proj.weight", "model.layers.31.mlp.experts.56.gate_and_up_proj.weight", "model.layers.31.mlp.experts.56.down_proj.weight", "model.layers.31.mlp.experts.57.gate_and_up_proj.weight", "model.layers.31.mlp.experts.57.down_proj.weight", "model.layers.31.mlp.experts.58.gate_and_up_proj.weight", "model.layers.31.mlp.experts.58.down_proj.weight", "model.layers.31.mlp.experts.59.gate_and_up_proj.weight", "model.layers.31.mlp.experts.59.down_proj.weight", "model.layers.31.mlp.experts.60.gate_and_up_proj.weight", "model.layers.31.mlp.experts.60.down_proj.weight", "model.layers.31.mlp.experts.61.gate_and_up_proj.weight", "model.layers.31.mlp.experts.61.down_proj.weight", "model.layers.31.mlp.experts.62.gate_and_up_proj.weight", "model.layers.31.mlp.experts.62.down_proj.weight", "model.layers.31.mlp.experts.63.gate_and_up_proj.weight", "model.layers.31.mlp.experts.63.down_proj.weight" ] }, "modules_group_size_dict": {}, "modules_svd_dict": {}, "modules_to_not_convert": [ "timestep_emb", ".vision_model.head", "time_embed_2", ".time_embed_2", ".vision_aligner", ".vae", ".vision_model.embeddings", "vae", ".final_layer", ".patch_embed", ".timestep_emb", "vision_aligner", "head", "wte", "final_layer", "patch_embed", "post_layernorm", "lm_head", ".model.ln_f", ".model.wte", ".lm_head", "time_embed", "embeddings", ".time_embed", ".vision_model.post_layernorm", "ln_f" ], "non_blocking": false, "quant_conv": false, "quant_method": "sdnq", "quantization_device": null, "quantized_matmul_dtype": null, "return_device": null, "sdnq_version": "0.1.5", "svd_rank": 32, "svd_steps": 8, "use_dynamic_quantization": true, "use_grad_ckpt": true, "use_quantized_matmul": false, "use_quantized_matmul_conv": false, "use_static_quantization": true, "use_stochastic_rounding": false, "use_svd": false, "weights_dtype": "uint4" }