{ "add_classification_head": false, "anyres_pooling_size": 2, "anyres_vit_max_image_size": null, "anyres_vit_two_views": false, "architectures": [ "HunyuanImage3ForCausalMM" ], "attention_bias": false, "attention_dropout": 0.0, "attention_head_dim": 128, "auto_map": { "AutoConfig": "configuration_hunyuan_image_3.HunyuanImage3Config", "AutoModel": "modeling_hunyuan_image_3.HunyuanImage3Model", "AutoModelForCausalLM": "modeling_hunyuan_image_3.HunyuanImage3ForCausalMM" }, "bos_token_id": 127958, "capacity_factor": 1.0, "cfg_distilled": false, "cla_share_factor": 2, "class_num": 0, "cond_image_type": "vae_vit", "cond_token_attn_type": "joint_full", "dense_list": [ 4096, 0 ], "dtype": "bfloat16", "eod_token_id": 3, "eos_token_id": 127957, "group_limited_greedy": false, "head_dim": 128, "hidden_act": "silu", "hidden_size": 4096, "im_end_id": 128001, "im_newline_id": 11, "im_start_id": 128000, "image_base_size": 1024, "image_token_id": 128006, "img_proj_type": "unet", "initializer_range": 0.02, "intermediate_size": 3072, "kv_lora_rank": null, "mask_init_id": 12, "max_position_embeddings": 22800, "mlp_bias": false, "model_type": "Hunyuan", "moe_drop_tokens": false, "moe_impl": "eager", "moe_intermediate_size": [ 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072, 3072 ], "moe_layer_num_skipped": 0, "moe_random_routing_dropped_token": false, "moe_topk": [ 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8 ], "n_group": false, "norm_topk_prob": true, "norm_type": "rms", "num_attention_heads": 32, "num_experts": 64, "num_hidden_layers": 32, "num_key_value_heads": 8, "num_media_embeds": 257, "num_shared_expert": [ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 ], "pad_id": 128009, "pad_token_id": 128009, "patch_embed_hidden_dim": 1024, "patch_size": 1, "pool_type": "last", "position_embedding_xdrope": false, "pretraining_tp": 1, "q_lora_rank": null, "qk_nope_head_dim": null, "qk_rope_head_dim": null, "quantization_config": { "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", "model.layers.0.mlp.experts.25.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.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": [ ".vision_aligner", ".vision_model.post_layernorm", "vision_aligner", ".timestep_emb", "vae", ".time_embed", ".time_embed_2", "ln_f", ".vision_model.embeddings", "lm_head", ".vae", ".model.wte", ".final_layer", "wte", ".vision_model.head", ".lm_head", ".model.ln_f", "timestep_emb", ".patch_embed", "time_embed_2", "post_layernorm", "time_embed", "patch_embed", "head", "embeddings", "final_layer" ], "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" }, "rms_norm_eps": 1e-05, "rope_parameters": { "alpha": 1.0, "beta_fast": 32, "beta_slow": 1, "factor": 1.0, "mscale": 1.0, "mscale_all_dim": 1.0, "rope_theta": 10000.0, "rope_type": "custom", "type": "custom" }, "rope_theta": 10000.0, "rope_type": "2d", "routed_scaling_factor": false, "skip_cls_token": false, "text_end_id": 7, "text_start_id": 6, "tie_word_embeddings": false, "topk_group": false, "transformers_version": "5.0.0", "use_cache": true, "use_cla": false, "use_meanflow": false, "use_mixed_mlp_moe": true, "use_mla": false, "use_qk_norm": true, "use_rotary_pos_emb": true, "v_head_dim": null, "vae": { "_class_name": "AutoencoderKLConv3D", "block_out_channels": [ 128, 256, 512, 1024, 1024 ], "downsample_match_channel": true, "ffactor_spatial": 16, "ffactor_temporal": 4, "in_channels": 3, "latent_channels": 32, "layers_per_block": 2, "out_channels": 3, "sample_size": 384, "sample_tsize": 96, "scaling_factor": 0.562679178327931, "upsample_match_channel": true }, "vae_autocast_dtype": "float16", "vae_downsample_factor": [ 16, 16 ], "vae_dtype": "float32", "vae_type": "hunyuan-image-vae-v1", "video_end_id": 10, "video_start_id": 9, "vit": { "_attn_implementation": "sdpa", "attention_dropout": 0.0, "hidden_act": "gelu_pytorch_tanh", "hidden_size": 1152, "intermediate_size": 4304, "layer_norm_eps": 1e-06, "num_attention_heads": 16, "num_channels": 3, "num_hidden_layers": 27, "num_patches": 256, "output_attentions": false, "output_hidden_states": false, "patch_size": 16, "torch_dtype": "float32", "use_return_dict": true }, "vit_add_patchemb_bias": false, "vit_aligner": { "depth": 2, "input_dim": 1152, "n_embed": 4096, "projector_type": "mlp_gelu", "torch_dtype": "float32" }, "vit_downsample_factor": [ 16, 16 ], "vit_input_resolution": 224, "vit_mapping_type": "resampler", "vit_norm_type": "fused", "vit_patch": 1, "vit_path": null, "vit_processor": { "do_convert_rgb": null, "do_normalize": true, "do_rescale": true, "do_resize": true, "image_mean": [ 0.5, 0.5, 0.5 ], "image_processor_type": "Siglip2ImageProcessorFast", "image_std": [ 0.5, 0.5, 0.5 ], "max_num_patches": 1024, "patch_size": 16, "processor_class": "Siglip2Processor", "resample": 2, "rescale_factor": 0.00392156862745098 }, "vit_remove_prenorm": false, "vit_token": 64, "vit_type": "siglip2-so400m-patch16-naflex", "vit_used_rms_norm": false, "vocab_size": 133120, "xdrope_section": null }