{
    "metadata": {
        "total_size": 162572866816
    },
    "weight_map": {
        "model.embed_tokens.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.input_layernorm.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.linear_attn.q_conv1d.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.linear_attn.k_conv1d.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.linear_attn.v_conv1d.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.linear_attn.q_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.linear_attn.k_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.linear_attn.v_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.linear_attn.f_a_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.linear_attn.f_b_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.linear_attn.b_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.linear_attn.g_a_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.linear_attn.g_b_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.linear_attn.o_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.linear_attn.a_log_bias": "model-00001-of-00049.safetensors",
        "model.layers.0.linear_attn.dt_bias": "model-00001-of-00049.safetensors",
        "model.layers.0.linear_attn.o_norm.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.mlp.gate.e_score_correction_bias": "model-00001-of-00049.safetensors",
        "model.layers.0.mlp.gate.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.mlp.shared_expert_gate.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.mlp.shared_expert.up_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.mlp.shared_expert.gate_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.mlp.shared_expert.down_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.mlp.experts.gate_up_proj": "model-00001-of-00049.safetensors",
        "model.layers.0.mlp.experts.down_proj": "model-00001-of-00049.safetensors",
        "model.layers.0.mlp_res_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.0.mlp_res_norm_weight": "model-00001-of-00049.safetensors",
        "model.layers.1.input_layernorm.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.linear_attn.q_conv1d.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.linear_attn.k_conv1d.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.linear_attn.v_conv1d.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.linear_attn.q_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.linear_attn.k_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.linear_attn.v_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.linear_attn.f_a_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.linear_attn.f_b_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.linear_attn.b_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.linear_attn.g_a_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.linear_attn.g_b_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.linear_attn.o_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.linear_attn.a_log_bias": "model-00001-of-00049.safetensors",
        "model.layers.1.linear_attn.dt_bias": "model-00001-of-00049.safetensors",
        "model.layers.1.linear_attn.o_norm.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.mlp.gate.e_score_correction_bias": "model-00001-of-00049.safetensors",
        "model.layers.1.mlp.gate.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.mlp.shared_expert_gate.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.mlp.shared_expert.up_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.mlp.shared_expert.gate_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.mlp.shared_expert.down_proj.weight": "model-00001-of-00049.safetensors",
        "model.layers.1.mlp.experts.gate_up_proj": "model-00002-of-00049.safetensors",
        "model.layers.1.mlp.experts.down_proj": "model-00002-of-00049.safetensors",
        "model.layers.1.attn_res_proj.weight": "model-00002-of-00049.safetensors",
        "model.layers.1.attn_res_norm_weight": "model-00002-of-00049.safetensors",
        "model.layers.1.mlp_res_proj.weight": "model-00002-of-00049.safetensors",
        "model.layers.1.mlp_res_norm_weight": "model-00002-of-00049.safetensors",
        "model.layers.2.input_layernorm.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.post_attention_layernorm.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.linear_attn.q_conv1d.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.linear_attn.k_conv1d.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.linear_attn.v_conv1d.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.linear_attn.q_proj.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.linear_attn.k_proj.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.linear_attn.v_proj.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.linear_attn.f_a_proj.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.linear_attn.f_b_proj.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.linear_attn.b_proj.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.linear_attn.g_a_proj.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.linear_attn.g_b_proj.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.linear_attn.o_proj.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.linear_attn.a_log_bias": "model-00002-of-00049.safetensors",
        "model.layers.2.linear_attn.dt_bias": "model-00002-of-00049.safetensors",
        "model.layers.2.linear_attn.o_norm.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.mlp.gate.e_score_correction_bias": "model-00002-of-00049.safetensors",
        "model.layers.2.mlp.gate.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.mlp.shared_expert_gate.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.mlp.shared_expert.up_proj.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.mlp.shared_expert.gate_proj.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.mlp.shared_expert.down_proj.weight": "model-00002-of-00049.safetensors",
        "model.layers.2.mlp.experts.gate_up_proj": "model-00003-of-00049.safetensors",
        "model.layers.2.mlp.experts.down_proj": "model-00003-of-00049.safetensors",
        "model.layers.2.attn_res_proj.weight": "model-00003-of-00049.safetensors",
        "model.layers.2.attn_res_norm_weight": "model-00003-of-00049.safetensors",
        "model.layers.2.mlp_res_proj.weight": "model-00003-of-00049.safetensors",
        "model.layers.2.mlp_res_norm_weight": "model-00003-of-00049.safetensors",
        "model.layers.3.input_layernorm.weight": "model-00003-of-00049.safetensors",
        "model.layers.3.post_attention_layernorm.weight": "model-00003-of-00049.safetensors",
        "model.layers.3.self_attn.q_proj.weight": "model-00003-of-00049.safetensors",
        "model.layers.3.self_attn.k_proj.weight": "model-00003-of-00049.safetensors",
        "model.layers.3.self_attn.v_proj.weight": "model-00003-of-00049.safetensors",
        "model.layers.3.self_attn.o_proj.weight": "model-00003-of-00049.safetensors",
        "model.layers.3.self_attn.k_norm.weight": "model-00003-of-00049.safetensors",
        "model.layers.3.self_attn.q_norm.weight": "model-00003-of-00049.safetensors",
        "model.layers.3.mlp.gate.e_score_correction_bias": "model-00003-of-00049.safetensors",
        "model.layers.3.mlp.gate.weight": "model-00003-of-00049.safetensors",
        "model.layers.3.mlp.shared_expert_gate.weight": "model-00003-of-00049.safetensors",
        "model.layers.3.mlp.shared_expert.up_proj.weight": "model-00003-of-00049.safetensors",
        "model.layers.3.mlp.shared_expert.gate_proj.weight": "model-00003-of-00049.safetensors",
        "model.layers.3.mlp.shared_expert.down_proj.weight": "model-00003-of-00049.safetensors",
        "model.layers.3.mlp.experts.gate_up_proj": "model-00004-of-00049.safetensors",
        "model.layers.3.mlp.experts.down_proj": "model-00004-of-00049.safetensors",
        "model.layers.3.attn_res_proj.weight": "model-00004-of-00049.safetensors",
        "model.layers.3.attn_res_norm_weight": "model-00004-of-00049.safetensors",
        "model.layers.3.mlp_res_proj.weight": "model-00004-of-00049.safetensors",
        "model.layers.3.mlp_res_norm_weight": "model-00004-of-00049.safetensors",
        "model.layers.4.input_layernorm.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.post_attention_layernorm.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.linear_attn.q_conv1d.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.linear_attn.k_conv1d.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.linear_attn.v_conv1d.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.linear_attn.q_proj.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.linear_attn.k_proj.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.linear_attn.v_proj.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.linear_attn.f_a_proj.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.linear_attn.f_b_proj.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.linear_attn.b_proj.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.linear_attn.g_a_proj.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.linear_attn.g_b_proj.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.linear_attn.o_proj.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.linear_attn.a_log_bias": "model-00004-of-00049.safetensors",
        "model.layers.4.linear_attn.dt_bias": "model-00004-of-00049.safetensors",
        "model.layers.4.linear_attn.o_norm.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.mlp.gate.e_score_correction_bias": "model-00004-of-00049.safetensors",
        "model.layers.4.mlp.gate.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.mlp.shared_expert_gate.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.mlp.shared_expert.up_proj.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.mlp.shared_expert.gate_proj.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.mlp.shared_expert.down_proj.weight": "model-00004-of-00049.safetensors",
        "model.layers.4.mlp.experts.gate_up_proj": "model-00005-of-00049.safetensors",
        "model.layers.4.mlp.experts.down_proj": "model-00005-of-00049.safetensors",
        "model.layers.4.attn_res_proj.weight": "model-00005-of-00049.safetensors",
        "model.layers.4.attn_res_norm_weight": "model-00005-of-00049.safetensors",
        "model.layers.4.mlp_res_proj.weight": "model-00005-of-00049.safetensors",
        "model.layers.4.mlp_res_norm_weight": "model-00005-of-00049.safetensors",
        "model.layers.5.input_layernorm.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.post_attention_layernorm.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.linear_attn.q_conv1d.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.linear_attn.k_conv1d.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.linear_attn.v_conv1d.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.linear_attn.q_proj.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.linear_attn.k_proj.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.linear_attn.v_proj.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.linear_attn.f_a_proj.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.linear_attn.f_b_proj.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.linear_attn.b_proj.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.linear_attn.g_a_proj.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.linear_attn.g_b_proj.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.linear_attn.o_proj.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.linear_attn.a_log_bias": "model-00005-of-00049.safetensors",
        "model.layers.5.linear_attn.dt_bias": "model-00005-of-00049.safetensors",
        "model.layers.5.linear_attn.o_norm.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.mlp.gate.e_score_correction_bias": "model-00005-of-00049.safetensors",
        "model.layers.5.mlp.gate.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.mlp.shared_expert_gate.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.mlp.shared_expert.up_proj.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.mlp.shared_expert.gate_proj.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.mlp.shared_expert.down_proj.weight": "model-00005-of-00049.safetensors",
        "model.layers.5.mlp.experts.gate_up_proj": "model-00006-of-00049.safetensors",
        "model.layers.5.mlp.experts.down_proj": "model-00006-of-00049.safetensors",
        "model.layers.5.attn_res_proj.weight": "model-00006-of-00049.safetensors",
        "model.layers.5.attn_res_norm_weight": "model-00006-of-00049.safetensors",
        "model.layers.5.mlp_res_proj.weight": "model-00006-of-00049.safetensors",
        "model.layers.5.mlp_res_norm_weight": "model-00006-of-00049.safetensors",
        "model.layers.6.input_layernorm.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.post_attention_layernorm.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.linear_attn.q_conv1d.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.linear_attn.k_conv1d.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.linear_attn.v_conv1d.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.linear_attn.q_proj.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.linear_attn.k_proj.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.linear_attn.v_proj.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.linear_attn.f_a_proj.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.linear_attn.f_b_proj.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.linear_attn.b_proj.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.linear_attn.g_a_proj.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.linear_attn.g_b_proj.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.linear_attn.o_proj.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.linear_attn.a_log_bias": "model-00006-of-00049.safetensors",
        "model.layers.6.linear_attn.dt_bias": "model-00006-of-00049.safetensors",
        "model.layers.6.linear_attn.o_norm.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.mlp.gate.e_score_correction_bias": "model-00006-of-00049.safetensors",
        "model.layers.6.mlp.gate.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.mlp.shared_expert_gate.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.mlp.shared_expert.up_proj.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.mlp.shared_expert.gate_proj.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.mlp.shared_expert.down_proj.weight": "model-00006-of-00049.safetensors",
        "model.layers.6.mlp.experts.gate_up_proj": "model-00007-of-00049.safetensors",
        "model.layers.6.mlp.experts.down_proj": "model-00007-of-00049.safetensors",
        "model.layers.6.attn_res_proj.weight": "model-00007-of-00049.safetensors",
        "model.layers.6.attn_res_norm_weight": "model-00007-of-00049.safetensors",
        "model.layers.6.mlp_res_proj.weight": "model-00007-of-00049.safetensors",
        "model.layers.6.mlp_res_norm_weight": "model-00007-of-00049.safetensors",
        "model.layers.7.input_layernorm.weight": "model-00007-of-00049.safetensors",
        "model.layers.7.post_attention_layernorm.weight": "model-00007-of-00049.safetensors",
        "model.layers.7.self_attn.q_proj.weight": "model-00007-of-00049.safetensors",
        "model.layers.7.self_attn.k_proj.weight": "model-00007-of-00049.safetensors",
        "model.layers.7.self_attn.v_proj.weight": "model-00007-of-00049.safetensors",
        "model.layers.7.self_attn.o_proj.weight": "model-00007-of-00049.safetensors",
        "model.layers.7.self_attn.k_norm.weight": "model-00007-of-00049.safetensors",
        "model.layers.7.self_attn.q_norm.weight": "model-00007-of-00049.safetensors",
        "model.layers.7.mlp.gate.e_score_correction_bias": "model-00007-of-00049.safetensors",
        "model.layers.7.mlp.gate.weight": "model-00007-of-00049.safetensors",
        "model.layers.7.mlp.shared_expert_gate.weight": "model-00007-of-00049.safetensors",
        "model.layers.7.mlp.shared_expert.up_proj.weight": "model-00007-of-00049.safetensors",
        "model.layers.7.mlp.shared_expert.gate_proj.weight": "model-00007-of-00049.safetensors",
        "model.layers.7.mlp.shared_expert.down_proj.weight": "model-00007-of-00049.safetensors",
        "model.layers.7.mlp.experts.gate_up_proj": "model-00008-of-00049.safetensors",
        "model.layers.7.mlp.experts.down_proj": "model-00008-of-00049.safetensors",
        "model.layers.7.attn_res_proj.weight": "model-00008-of-00049.safetensors",
        "model.layers.7.attn_res_norm_weight": "model-00008-of-00049.safetensors",
        "model.layers.7.mlp_res_proj.weight": "model-00008-of-00049.safetensors",
        "model.layers.7.mlp_res_norm_weight": "model-00008-of-00049.safetensors",
        "model.layers.8.input_layernorm.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.post_attention_layernorm.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.linear_attn.q_conv1d.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.linear_attn.k_conv1d.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.linear_attn.v_conv1d.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.linear_attn.q_proj.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.linear_attn.k_proj.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.linear_attn.v_proj.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.linear_attn.f_a_proj.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.linear_attn.f_b_proj.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.linear_attn.b_proj.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.linear_attn.g_a_proj.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.linear_attn.g_b_proj.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.linear_attn.o_proj.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.linear_attn.a_log_bias": "model-00008-of-00049.safetensors",
        "model.layers.8.linear_attn.dt_bias": "model-00008-of-00049.safetensors",
        "model.layers.8.linear_attn.o_norm.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.mlp.gate.e_score_correction_bias": "model-00008-of-00049.safetensors",
        "model.layers.8.mlp.gate.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.mlp.shared_expert_gate.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.mlp.shared_expert.up_proj.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.mlp.shared_expert.gate_proj.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.mlp.shared_expert.down_proj.weight": "model-00008-of-00049.safetensors",
        "model.layers.8.mlp.experts.gate_up_proj": "model-00009-of-00049.safetensors",
        "model.layers.8.mlp.experts.down_proj": "model-00009-of-00049.safetensors",
        "model.layers.8.attn_res_proj.weight": "model-00009-of-00049.safetensors",
        "model.layers.8.attn_res_norm_weight": "model-00009-of-00049.safetensors",
        "model.layers.8.mlp_res_proj.weight": "model-00009-of-00049.safetensors",
        "model.layers.8.mlp_res_norm_weight": "model-00009-of-00049.safetensors",
        "model.layers.9.input_layernorm.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.post_attention_layernorm.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.linear_attn.q_conv1d.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.linear_attn.k_conv1d.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.linear_attn.v_conv1d.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.linear_attn.q_proj.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.linear_attn.k_proj.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.linear_attn.v_proj.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.linear_attn.f_a_proj.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.linear_attn.f_b_proj.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.linear_attn.b_proj.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.linear_attn.g_a_proj.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.linear_attn.g_b_proj.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.linear_attn.o_proj.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.linear_attn.a_log_bias": "model-00009-of-00049.safetensors",
        "model.layers.9.linear_attn.dt_bias": "model-00009-of-00049.safetensors",
        "model.layers.9.linear_attn.o_norm.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.mlp.gate.e_score_correction_bias": "model-00009-of-00049.safetensors",
        "model.layers.9.mlp.gate.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.mlp.shared_expert_gate.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.mlp.shared_expert.up_proj.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.mlp.shared_expert.gate_proj.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.mlp.shared_expert.down_proj.weight": "model-00009-of-00049.safetensors",
        "model.layers.9.mlp.experts.gate_up_proj": "model-00010-of-00049.safetensors",
        "model.layers.9.mlp.experts.down_proj": "model-00010-of-00049.safetensors",
        "model.layers.9.attn_res_proj.weight": "model-00010-of-00049.safetensors",
        "model.layers.9.attn_res_norm_weight": "model-00010-of-00049.safetensors",
        "model.layers.9.mlp_res_proj.weight": "model-00010-of-00049.safetensors",
        "model.layers.9.mlp_res_norm_weight": "model-00010-of-00049.safetensors",
        "model.layers.10.input_layernorm.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.post_attention_layernorm.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.linear_attn.q_conv1d.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.linear_attn.k_conv1d.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.linear_attn.v_conv1d.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.linear_attn.q_proj.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.linear_attn.k_proj.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.linear_attn.v_proj.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.linear_attn.f_a_proj.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.linear_attn.f_b_proj.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.linear_attn.b_proj.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.linear_attn.g_a_proj.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.linear_attn.g_b_proj.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.linear_attn.o_proj.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.linear_attn.a_log_bias": "model-00010-of-00049.safetensors",
        "model.layers.10.linear_attn.dt_bias": "model-00010-of-00049.safetensors",
        "model.layers.10.linear_attn.o_norm.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.mlp.gate.e_score_correction_bias": "model-00010-of-00049.safetensors",
        "model.layers.10.mlp.gate.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.mlp.shared_expert_gate.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.mlp.shared_expert.up_proj.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.mlp.shared_expert.gate_proj.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.mlp.shared_expert.down_proj.weight": "model-00010-of-00049.safetensors",
        "model.layers.10.mlp.experts.gate_up_proj": "model-00011-of-00049.safetensors",
        "model.layers.10.mlp.experts.down_proj": "model-00011-of-00049.safetensors",
        "model.layers.10.attn_res_proj.weight": "model-00011-of-00049.safetensors",
        "model.layers.10.attn_res_norm_weight": "model-00011-of-00049.safetensors",
        "model.layers.10.mlp_res_proj.weight": "model-00011-of-00049.safetensors",
        "model.layers.10.mlp_res_norm_weight": "model-00011-of-00049.safetensors",
        "model.layers.11.input_layernorm.weight": "model-00011-of-00049.safetensors",
        "model.layers.11.post_attention_layernorm.weight": "model-00011-of-00049.safetensors",
        "model.layers.11.self_attn.q_proj.weight": "model-00011-of-00049.safetensors",
        "model.layers.11.self_attn.k_proj.weight": "model-00011-of-00049.safetensors",
        "model.layers.11.self_attn.v_proj.weight": "model-00011-of-00049.safetensors",
        "model.layers.11.self_attn.o_proj.weight": "model-00011-of-00049.safetensors",
        "model.layers.11.self_attn.k_norm.weight": "model-00011-of-00049.safetensors",
        "model.layers.11.self_attn.q_norm.weight": "model-00011-of-00049.safetensors",
        "model.layers.11.mlp.gate.e_score_correction_bias": "model-00011-of-00049.safetensors",
        "model.layers.11.mlp.gate.weight": "model-00011-of-00049.safetensors",
        "model.layers.11.mlp.shared_expert_gate.weight": "model-00011-of-00049.safetensors",
        "model.layers.11.mlp.shared_expert.up_proj.weight": "model-00011-of-00049.safetensors",
        "model.layers.11.mlp.shared_expert.gate_proj.weight": "model-00011-of-00049.safetensors",
        "model.layers.11.mlp.shared_expert.down_proj.weight": "model-00011-of-00049.safetensors",
        "model.layers.11.mlp.experts.gate_up_proj": "model-00012-of-00049.safetensors",
        "model.layers.11.mlp.experts.down_proj": "model-00012-of-00049.safetensors",
        "model.layers.11.attn_res_proj.weight": "model-00012-of-00049.safetensors",
        "model.layers.11.attn_res_norm_weight": "model-00012-of-00049.safetensors",
        "model.layers.11.mlp_res_proj.weight": "model-00012-of-00049.safetensors",
        "model.layers.11.mlp_res_norm_weight": "model-00012-of-00049.safetensors",
        "model.layers.12.input_layernorm.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.post_attention_layernorm.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.linear_attn.q_conv1d.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.linear_attn.k_conv1d.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.linear_attn.v_conv1d.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.linear_attn.q_proj.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.linear_attn.k_proj.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.linear_attn.v_proj.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.linear_attn.f_a_proj.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.linear_attn.f_b_proj.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.linear_attn.b_proj.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.linear_attn.g_a_proj.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.linear_attn.g_b_proj.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.linear_attn.o_proj.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.linear_attn.a_log_bias": "model-00012-of-00049.safetensors",
        "model.layers.12.linear_attn.dt_bias": "model-00012-of-00049.safetensors",
        "model.layers.12.linear_attn.o_norm.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.mlp.gate.e_score_correction_bias": "model-00012-of-00049.safetensors",
        "model.layers.12.mlp.gate.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.mlp.shared_expert_gate.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.mlp.shared_expert.up_proj.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.mlp.shared_expert.gate_proj.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.mlp.shared_expert.down_proj.weight": "model-00012-of-00049.safetensors",
        "model.layers.12.mlp.experts.gate_up_proj": "model-00013-of-00049.safetensors",
        "model.layers.12.mlp.experts.down_proj": "model-00013-of-00049.safetensors",
        "model.layers.12.attn_res_proj.weight": "model-00013-of-00049.safetensors",
        "model.layers.12.attn_res_norm_weight": "model-00013-of-00049.safetensors",
        "model.layers.12.mlp_res_proj.weight": "model-00013-of-00049.safetensors",
        "model.layers.12.mlp_res_norm_weight": "model-00013-of-00049.safetensors",
        "model.layers.13.input_layernorm.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.post_attention_layernorm.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.linear_attn.q_conv1d.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.linear_attn.k_conv1d.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.linear_attn.v_conv1d.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.linear_attn.q_proj.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.linear_attn.k_proj.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.linear_attn.v_proj.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.linear_attn.f_a_proj.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.linear_attn.f_b_proj.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.linear_attn.b_proj.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.linear_attn.g_a_proj.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.linear_attn.g_b_proj.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.linear_attn.o_proj.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.linear_attn.a_log_bias": "model-00013-of-00049.safetensors",
        "model.layers.13.linear_attn.dt_bias": "model-00013-of-00049.safetensors",
        "model.layers.13.linear_attn.o_norm.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.mlp.gate.e_score_correction_bias": "model-00013-of-00049.safetensors",
        "model.layers.13.mlp.gate.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.mlp.shared_expert_gate.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.mlp.shared_expert.up_proj.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.mlp.shared_expert.gate_proj.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.mlp.shared_expert.down_proj.weight": "model-00013-of-00049.safetensors",
        "model.layers.13.mlp.experts.gate_up_proj": "model-00014-of-00049.safetensors",
        "model.layers.13.mlp.experts.down_proj": "model-00014-of-00049.safetensors",
        "model.layers.13.attn_res_proj.weight": "model-00014-of-00049.safetensors",
        "model.layers.13.attn_res_norm_weight": "model-00014-of-00049.safetensors",
        "model.layers.13.mlp_res_proj.weight": "model-00014-of-00049.safetensors",
        "model.layers.13.mlp_res_norm_weight": "model-00014-of-00049.safetensors",
        "model.layers.14.input_layernorm.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.post_attention_layernorm.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.linear_attn.q_conv1d.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.linear_attn.k_conv1d.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.linear_attn.v_conv1d.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.linear_attn.q_proj.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.linear_attn.k_proj.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.linear_attn.v_proj.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.linear_attn.f_a_proj.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.linear_attn.f_b_proj.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.linear_attn.b_proj.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.linear_attn.g_a_proj.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.linear_attn.g_b_proj.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.linear_attn.o_proj.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.linear_attn.a_log_bias": "model-00014-of-00049.safetensors",
        "model.layers.14.linear_attn.dt_bias": "model-00014-of-00049.safetensors",
        "model.layers.14.linear_attn.o_norm.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.mlp.gate.e_score_correction_bias": "model-00014-of-00049.safetensors",
        "model.layers.14.mlp.gate.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.mlp.shared_expert_gate.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.mlp.shared_expert.up_proj.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.mlp.shared_expert.gate_proj.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.mlp.shared_expert.down_proj.weight": "model-00014-of-00049.safetensors",
        "model.layers.14.mlp.experts.gate_up_proj": "model-00015-of-00049.safetensors",
        "model.layers.14.mlp.experts.down_proj": "model-00015-of-00049.safetensors",
        "model.layers.14.attn_res_proj.weight": "model-00015-of-00049.safetensors",
        "model.layers.14.attn_res_norm_weight": "model-00015-of-00049.safetensors",
        "model.layers.14.mlp_res_proj.weight": "model-00015-of-00049.safetensors",
        "model.layers.14.mlp_res_norm_weight": "model-00015-of-00049.safetensors",
        "model.layers.15.input_layernorm.weight": "model-00015-of-00049.safetensors",
        "model.layers.15.post_attention_layernorm.weight": "model-00015-of-00049.safetensors",
        "model.layers.15.self_attn.q_proj.weight": "model-00015-of-00049.safetensors",
        "model.layers.15.self_attn.k_proj.weight": "model-00015-of-00049.safetensors",
        "model.layers.15.self_attn.v_proj.weight": "model-00015-of-00049.safetensors",
        "model.layers.15.self_attn.o_proj.weight": "model-00015-of-00049.safetensors",
        "model.layers.15.self_attn.k_norm.weight": "model-00015-of-00049.safetensors",
        "model.layers.15.self_attn.q_norm.weight": "model-00015-of-00049.safetensors",
        "model.layers.15.mlp.gate.e_score_correction_bias": "model-00015-of-00049.safetensors",
        "model.layers.15.mlp.gate.weight": "model-00015-of-00049.safetensors",
        "model.layers.15.mlp.shared_expert_gate.weight": "model-00015-of-00049.safetensors",
        "model.layers.15.mlp.shared_expert.up_proj.weight": "model-00015-of-00049.safetensors",
        "model.layers.15.mlp.shared_expert.gate_proj.weight": "model-00015-of-00049.safetensors",
        "model.layers.15.mlp.shared_expert.down_proj.weight": "model-00015-of-00049.safetensors",
        "model.layers.15.mlp.experts.gate_up_proj": "model-00016-of-00049.safetensors",
        "model.layers.15.mlp.experts.down_proj": "model-00016-of-00049.safetensors",
        "model.layers.15.attn_res_proj.weight": "model-00016-of-00049.safetensors",
        "model.layers.15.attn_res_norm_weight": "model-00016-of-00049.safetensors",
        "model.layers.15.mlp_res_proj.weight": "model-00016-of-00049.safetensors",
        "model.layers.15.mlp_res_norm_weight": "model-00016-of-00049.safetensors",
        "model.layers.16.input_layernorm.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.post_attention_layernorm.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.linear_attn.q_conv1d.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.linear_attn.k_conv1d.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.linear_attn.v_conv1d.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.linear_attn.q_proj.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.linear_attn.k_proj.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.linear_attn.v_proj.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.linear_attn.f_a_proj.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.linear_attn.f_b_proj.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.linear_attn.b_proj.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.linear_attn.g_a_proj.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.linear_attn.g_b_proj.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.linear_attn.o_proj.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.linear_attn.a_log_bias": "model-00016-of-00049.safetensors",
        "model.layers.16.linear_attn.dt_bias": "model-00016-of-00049.safetensors",
        "model.layers.16.linear_attn.o_norm.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.mlp.gate.e_score_correction_bias": "model-00016-of-00049.safetensors",
        "model.layers.16.mlp.gate.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.mlp.shared_expert_gate.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.mlp.shared_expert.up_proj.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.mlp.shared_expert.gate_proj.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.mlp.shared_expert.down_proj.weight": "model-00016-of-00049.safetensors",
        "model.layers.16.mlp.experts.gate_up_proj": "model-00017-of-00049.safetensors",
        "model.layers.16.mlp.experts.down_proj": "model-00017-of-00049.safetensors",
        "model.layers.16.attn_res_proj.weight": "model-00017-of-00049.safetensors",
        "model.layers.16.attn_res_norm_weight": "model-00017-of-00049.safetensors",
        "model.layers.16.mlp_res_proj.weight": "model-00017-of-00049.safetensors",
        "model.layers.16.mlp_res_norm_weight": "model-00017-of-00049.safetensors",
        "model.layers.17.input_layernorm.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.post_attention_layernorm.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.linear_attn.q_conv1d.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.linear_attn.k_conv1d.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.linear_attn.v_conv1d.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.linear_attn.q_proj.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.linear_attn.k_proj.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.linear_attn.v_proj.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.linear_attn.f_a_proj.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.linear_attn.f_b_proj.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.linear_attn.b_proj.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.linear_attn.g_a_proj.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.linear_attn.g_b_proj.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.linear_attn.o_proj.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.linear_attn.a_log_bias": "model-00017-of-00049.safetensors",
        "model.layers.17.linear_attn.dt_bias": "model-00017-of-00049.safetensors",
        "model.layers.17.linear_attn.o_norm.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.mlp.gate.e_score_correction_bias": "model-00017-of-00049.safetensors",
        "model.layers.17.mlp.gate.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.mlp.shared_expert_gate.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.mlp.shared_expert.up_proj.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.mlp.shared_expert.gate_proj.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.mlp.shared_expert.down_proj.weight": "model-00017-of-00049.safetensors",
        "model.layers.17.mlp.experts.gate_up_proj": "model-00018-of-00049.safetensors",
        "model.layers.17.mlp.experts.down_proj": "model-00018-of-00049.safetensors",
        "model.layers.17.attn_res_proj.weight": "model-00018-of-00049.safetensors",
        "model.layers.17.attn_res_norm_weight": "model-00018-of-00049.safetensors",
        "model.layers.17.mlp_res_proj.weight": "model-00018-of-00049.safetensors",
        "model.layers.17.mlp_res_norm_weight": "model-00018-of-00049.safetensors",
        "model.layers.18.input_layernorm.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.post_attention_layernorm.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.linear_attn.q_conv1d.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.linear_attn.k_conv1d.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.linear_attn.v_conv1d.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.linear_attn.q_proj.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.linear_attn.k_proj.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.linear_attn.v_proj.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.linear_attn.f_a_proj.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.linear_attn.f_b_proj.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.linear_attn.b_proj.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.linear_attn.g_a_proj.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.linear_attn.g_b_proj.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.linear_attn.o_proj.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.linear_attn.a_log_bias": "model-00018-of-00049.safetensors",
        "model.layers.18.linear_attn.dt_bias": "model-00018-of-00049.safetensors",
        "model.layers.18.linear_attn.o_norm.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.mlp.gate.e_score_correction_bias": "model-00018-of-00049.safetensors",
        "model.layers.18.mlp.gate.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.mlp.shared_expert_gate.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.mlp.shared_expert.up_proj.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.mlp.shared_expert.gate_proj.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.mlp.shared_expert.down_proj.weight": "model-00018-of-00049.safetensors",
        "model.layers.18.mlp.experts.gate_up_proj": "model-00019-of-00049.safetensors",
        "model.layers.18.mlp.experts.down_proj": "model-00019-of-00049.safetensors",
        "model.layers.18.attn_res_proj.weight": "model-00019-of-00049.safetensors",
        "model.layers.18.attn_res_norm_weight": "model-00019-of-00049.safetensors",
        "model.layers.18.mlp_res_proj.weight": "model-00019-of-00049.safetensors",
        "model.layers.18.mlp_res_norm_weight": "model-00019-of-00049.safetensors",
        "model.layers.19.input_layernorm.weight": "model-00019-of-00049.safetensors",
        "model.layers.19.post_attention_layernorm.weight": "model-00019-of-00049.safetensors",
        "model.layers.19.self_attn.q_proj.weight": "model-00019-of-00049.safetensors",
        "model.layers.19.self_attn.k_proj.weight": "model-00019-of-00049.safetensors",
        "model.layers.19.self_attn.v_proj.weight": "model-00019-of-00049.safetensors",
        "model.layers.19.self_attn.o_proj.weight": "model-00019-of-00049.safetensors",
        "model.layers.19.self_attn.k_norm.weight": "model-00019-of-00049.safetensors",
        "model.layers.19.self_attn.q_norm.weight": "model-00019-of-00049.safetensors",
        "model.layers.19.mlp.gate.e_score_correction_bias": "model-00019-of-00049.safetensors",
        "model.layers.19.mlp.gate.weight": "model-00019-of-00049.safetensors",
        "model.layers.19.mlp.shared_expert_gate.weight": "model-00019-of-00049.safetensors",
        "model.layers.19.mlp.shared_expert.up_proj.weight": "model-00019-of-00049.safetensors",
        "model.layers.19.mlp.shared_expert.gate_proj.weight": "model-00019-of-00049.safetensors",
        "model.layers.19.mlp.shared_expert.down_proj.weight": "model-00019-of-00049.safetensors",
        "model.layers.19.mlp.experts.gate_up_proj": "model-00020-of-00049.safetensors",
        "model.layers.19.mlp.experts.down_proj": "model-00020-of-00049.safetensors",
        "model.layers.19.attn_res_proj.weight": "model-00020-of-00049.safetensors",
        "model.layers.19.attn_res_norm_weight": "model-00020-of-00049.safetensors",
        "model.layers.19.mlp_res_proj.weight": "model-00020-of-00049.safetensors",
        "model.layers.19.mlp_res_norm_weight": "model-00020-of-00049.safetensors",
        "model.layers.20.input_layernorm.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.post_attention_layernorm.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.linear_attn.q_conv1d.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.linear_attn.k_conv1d.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.linear_attn.v_conv1d.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.linear_attn.q_proj.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.linear_attn.k_proj.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.linear_attn.v_proj.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.linear_attn.f_a_proj.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.linear_attn.f_b_proj.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.linear_attn.b_proj.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.linear_attn.g_a_proj.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.linear_attn.g_b_proj.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.linear_attn.o_proj.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.linear_attn.a_log_bias": "model-00020-of-00049.safetensors",
        "model.layers.20.linear_attn.dt_bias": "model-00020-of-00049.safetensors",
        "model.layers.20.linear_attn.o_norm.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.mlp.gate.e_score_correction_bias": "model-00020-of-00049.safetensors",
        "model.layers.20.mlp.gate.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.mlp.shared_expert_gate.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.mlp.shared_expert.up_proj.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.mlp.shared_expert.gate_proj.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.mlp.shared_expert.down_proj.weight": "model-00020-of-00049.safetensors",
        "model.layers.20.mlp.experts.gate_up_proj": "model-00021-of-00049.safetensors",
        "model.layers.20.mlp.experts.down_proj": "model-00021-of-00049.safetensors",
        "model.layers.20.attn_res_proj.weight": "model-00021-of-00049.safetensors",
        "model.layers.20.attn_res_norm_weight": "model-00021-of-00049.safetensors",
        "model.layers.20.mlp_res_proj.weight": "model-00021-of-00049.safetensors",
        "model.layers.20.mlp_res_norm_weight": "model-00021-of-00049.safetensors",
        "model.layers.21.input_layernorm.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.post_attention_layernorm.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.linear_attn.q_conv1d.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.linear_attn.k_conv1d.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.linear_attn.v_conv1d.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.linear_attn.q_proj.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.linear_attn.k_proj.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.linear_attn.v_proj.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.linear_attn.f_a_proj.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.linear_attn.f_b_proj.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.linear_attn.b_proj.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.linear_attn.g_a_proj.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.linear_attn.g_b_proj.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.linear_attn.o_proj.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.linear_attn.a_log_bias": "model-00021-of-00049.safetensors",
        "model.layers.21.linear_attn.dt_bias": "model-00021-of-00049.safetensors",
        "model.layers.21.linear_attn.o_norm.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.mlp.gate.e_score_correction_bias": "model-00021-of-00049.safetensors",
        "model.layers.21.mlp.gate.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.mlp.shared_expert_gate.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.mlp.shared_expert.up_proj.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.mlp.shared_expert.gate_proj.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.mlp.shared_expert.down_proj.weight": "model-00021-of-00049.safetensors",
        "model.layers.21.mlp.experts.gate_up_proj": "model-00022-of-00049.safetensors",
        "model.layers.21.mlp.experts.down_proj": "model-00022-of-00049.safetensors",
        "model.layers.21.attn_res_proj.weight": "model-00022-of-00049.safetensors",
        "model.layers.21.attn_res_norm_weight": "model-00022-of-00049.safetensors",
        "model.layers.21.mlp_res_proj.weight": "model-00022-of-00049.safetensors",
        "model.layers.21.mlp_res_norm_weight": "model-00022-of-00049.safetensors",
        "model.layers.22.input_layernorm.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.post_attention_layernorm.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.linear_attn.q_conv1d.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.linear_attn.k_conv1d.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.linear_attn.v_conv1d.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.linear_attn.q_proj.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.linear_attn.k_proj.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.linear_attn.v_proj.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.linear_attn.f_a_proj.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.linear_attn.f_b_proj.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.linear_attn.b_proj.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.linear_attn.g_a_proj.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.linear_attn.g_b_proj.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.linear_attn.o_proj.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.linear_attn.a_log_bias": "model-00022-of-00049.safetensors",
        "model.layers.22.linear_attn.dt_bias": "model-00022-of-00049.safetensors",
        "model.layers.22.linear_attn.o_norm.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.mlp.gate.e_score_correction_bias": "model-00022-of-00049.safetensors",
        "model.layers.22.mlp.gate.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.mlp.shared_expert_gate.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.mlp.shared_expert.up_proj.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.mlp.shared_expert.gate_proj.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.mlp.shared_expert.down_proj.weight": "model-00022-of-00049.safetensors",
        "model.layers.22.mlp.experts.gate_up_proj": "model-00023-of-00049.safetensors",
        "model.layers.22.mlp.experts.down_proj": "model-00023-of-00049.safetensors",
        "model.layers.22.attn_res_proj.weight": "model-00023-of-00049.safetensors",
        "model.layers.22.attn_res_norm_weight": "model-00023-of-00049.safetensors",
        "model.layers.22.mlp_res_proj.weight": "model-00023-of-00049.safetensors",
        "model.layers.22.mlp_res_norm_weight": "model-00023-of-00049.safetensors",
        "model.layers.23.input_layernorm.weight": "model-00023-of-00049.safetensors",
        "model.layers.23.post_attention_layernorm.weight": "model-00023-of-00049.safetensors",
        "model.layers.23.self_attn.q_proj.weight": "model-00023-of-00049.safetensors",
        "model.layers.23.self_attn.k_proj.weight": "model-00023-of-00049.safetensors",
        "model.layers.23.self_attn.v_proj.weight": "model-00023-of-00049.safetensors",
        "model.layers.23.self_attn.o_proj.weight": "model-00023-of-00049.safetensors",
        "model.layers.23.self_attn.k_norm.weight": "model-00023-of-00049.safetensors",
        "model.layers.23.self_attn.q_norm.weight": "model-00023-of-00049.safetensors",
        "model.layers.23.mlp.gate.e_score_correction_bias": "model-00023-of-00049.safetensors",
        "model.layers.23.mlp.gate.weight": "model-00023-of-00049.safetensors",
        "model.layers.23.mlp.shared_expert_gate.weight": "model-00023-of-00049.safetensors",
        "model.layers.23.mlp.shared_expert.up_proj.weight": "model-00023-of-00049.safetensors",
        "model.layers.23.mlp.shared_expert.gate_proj.weight": "model-00023-of-00049.safetensors",
        "model.layers.23.mlp.shared_expert.down_proj.weight": "model-00023-of-00049.safetensors",
        "model.layers.23.mlp.experts.gate_up_proj": "model-00024-of-00049.safetensors",
        "model.layers.23.mlp.experts.down_proj": "model-00024-of-00049.safetensors",
        "model.layers.23.attn_res_proj.weight": "model-00024-of-00049.safetensors",
        "model.layers.23.attn_res_norm_weight": "model-00024-of-00049.safetensors",
        "model.layers.23.mlp_res_proj.weight": "model-00024-of-00049.safetensors",
        "model.layers.23.mlp_res_norm_weight": "model-00024-of-00049.safetensors",
        "model.layers.24.input_layernorm.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.post_attention_layernorm.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.linear_attn.q_conv1d.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.linear_attn.k_conv1d.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.linear_attn.v_conv1d.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.linear_attn.q_proj.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.linear_attn.k_proj.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.linear_attn.v_proj.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.linear_attn.f_a_proj.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.linear_attn.f_b_proj.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.linear_attn.b_proj.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.linear_attn.g_a_proj.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.linear_attn.g_b_proj.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.linear_attn.o_proj.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.linear_attn.a_log_bias": "model-00024-of-00049.safetensors",
        "model.layers.24.linear_attn.dt_bias": "model-00024-of-00049.safetensors",
        "model.layers.24.linear_attn.o_norm.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.mlp.gate.e_score_correction_bias": "model-00024-of-00049.safetensors",
        "model.layers.24.mlp.gate.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.mlp.shared_expert_gate.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.mlp.shared_expert.up_proj.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.mlp.shared_expert.gate_proj.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.mlp.shared_expert.down_proj.weight": "model-00024-of-00049.safetensors",
        "model.layers.24.mlp.experts.gate_up_proj": "model-00025-of-00049.safetensors",
        "model.layers.24.mlp.experts.down_proj": "model-00025-of-00049.safetensors",
        "model.layers.24.attn_res_proj.weight": "model-00025-of-00049.safetensors",
        "model.layers.24.attn_res_norm_weight": "model-00025-of-00049.safetensors",
        "model.layers.24.mlp_res_proj.weight": "model-00025-of-00049.safetensors",
        "model.layers.24.mlp_res_norm_weight": "model-00025-of-00049.safetensors",
        "model.layers.25.input_layernorm.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.post_attention_layernorm.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.linear_attn.q_conv1d.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.linear_attn.k_conv1d.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.linear_attn.v_conv1d.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.linear_attn.q_proj.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.linear_attn.k_proj.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.linear_attn.v_proj.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.linear_attn.f_a_proj.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.linear_attn.f_b_proj.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.linear_attn.b_proj.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.linear_attn.g_a_proj.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.linear_attn.g_b_proj.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.linear_attn.o_proj.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.linear_attn.a_log_bias": "model-00025-of-00049.safetensors",
        "model.layers.25.linear_attn.dt_bias": "model-00025-of-00049.safetensors",
        "model.layers.25.linear_attn.o_norm.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.mlp.gate.e_score_correction_bias": "model-00025-of-00049.safetensors",
        "model.layers.25.mlp.gate.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.mlp.shared_expert_gate.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.mlp.shared_expert.up_proj.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.mlp.shared_expert.gate_proj.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.mlp.shared_expert.down_proj.weight": "model-00025-of-00049.safetensors",
        "model.layers.25.mlp.experts.gate_up_proj": "model-00026-of-00049.safetensors",
        "model.layers.25.mlp.experts.down_proj": "model-00026-of-00049.safetensors",
        "model.layers.25.attn_res_proj.weight": "model-00026-of-00049.safetensors",
        "model.layers.25.attn_res_norm_weight": "model-00026-of-00049.safetensors",
        "model.layers.25.mlp_res_proj.weight": "model-00026-of-00049.safetensors",
        "model.layers.25.mlp_res_norm_weight": "model-00026-of-00049.safetensors",
        "model.layers.26.input_layernorm.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.post_attention_layernorm.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.linear_attn.q_conv1d.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.linear_attn.k_conv1d.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.linear_attn.v_conv1d.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.linear_attn.q_proj.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.linear_attn.k_proj.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.linear_attn.v_proj.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.linear_attn.f_a_proj.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.linear_attn.f_b_proj.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.linear_attn.b_proj.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.linear_attn.g_a_proj.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.linear_attn.g_b_proj.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.linear_attn.o_proj.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.linear_attn.a_log_bias": "model-00026-of-00049.safetensors",
        "model.layers.26.linear_attn.dt_bias": "model-00026-of-00049.safetensors",
        "model.layers.26.linear_attn.o_norm.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.mlp.gate.e_score_correction_bias": "model-00026-of-00049.safetensors",
        "model.layers.26.mlp.gate.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.mlp.shared_expert_gate.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.mlp.shared_expert.up_proj.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.mlp.shared_expert.gate_proj.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.mlp.shared_expert.down_proj.weight": "model-00026-of-00049.safetensors",
        "model.layers.26.mlp.experts.gate_up_proj": "model-00027-of-00049.safetensors",
        "model.layers.26.mlp.experts.down_proj": "model-00027-of-00049.safetensors",
        "model.layers.26.attn_res_proj.weight": "model-00027-of-00049.safetensors",
        "model.layers.26.attn_res_norm_weight": "model-00027-of-00049.safetensors",
        "model.layers.26.mlp_res_proj.weight": "model-00027-of-00049.safetensors",
        "model.layers.26.mlp_res_norm_weight": "model-00027-of-00049.safetensors",
        "model.layers.27.input_layernorm.weight": "model-00027-of-00049.safetensors",
        "model.layers.27.post_attention_layernorm.weight": "model-00027-of-00049.safetensors",
        "model.layers.27.self_attn.q_proj.weight": "model-00027-of-00049.safetensors",
        "model.layers.27.self_attn.k_proj.weight": "model-00027-of-00049.safetensors",
        "model.layers.27.self_attn.v_proj.weight": "model-00027-of-00049.safetensors",
        "model.layers.27.self_attn.o_proj.weight": "model-00027-of-00049.safetensors",
        "model.layers.27.self_attn.k_norm.weight": "model-00027-of-00049.safetensors",
        "model.layers.27.self_attn.q_norm.weight": "model-00027-of-00049.safetensors",
        "model.layers.27.mlp.gate.e_score_correction_bias": "model-00027-of-00049.safetensors",
        "model.layers.27.mlp.gate.weight": "model-00027-of-00049.safetensors",
        "model.layers.27.mlp.shared_expert_gate.weight": "model-00027-of-00049.safetensors",
        "model.layers.27.mlp.shared_expert.up_proj.weight": "model-00027-of-00049.safetensors",
        "model.layers.27.mlp.shared_expert.gate_proj.weight": "model-00027-of-00049.safetensors",
        "model.layers.27.mlp.shared_expert.down_proj.weight": "model-00027-of-00049.safetensors",
        "model.layers.27.mlp.experts.gate_up_proj": "model-00028-of-00049.safetensors",
        "model.layers.27.mlp.experts.down_proj": "model-00028-of-00049.safetensors",
        "model.layers.27.attn_res_proj.weight": "model-00028-of-00049.safetensors",
        "model.layers.27.attn_res_norm_weight": "model-00028-of-00049.safetensors",
        "model.layers.27.mlp_res_proj.weight": "model-00028-of-00049.safetensors",
        "model.layers.27.mlp_res_norm_weight": "model-00028-of-00049.safetensors",
        "model.layers.28.input_layernorm.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.post_attention_layernorm.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.linear_attn.q_conv1d.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.linear_attn.k_conv1d.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.linear_attn.v_conv1d.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.linear_attn.q_proj.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.linear_attn.k_proj.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.linear_attn.v_proj.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.linear_attn.f_a_proj.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.linear_attn.f_b_proj.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.linear_attn.b_proj.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.linear_attn.g_a_proj.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.linear_attn.g_b_proj.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.linear_attn.o_proj.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.linear_attn.a_log_bias": "model-00028-of-00049.safetensors",
        "model.layers.28.linear_attn.dt_bias": "model-00028-of-00049.safetensors",
        "model.layers.28.linear_attn.o_norm.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.mlp.gate.e_score_correction_bias": "model-00028-of-00049.safetensors",
        "model.layers.28.mlp.gate.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.mlp.shared_expert_gate.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.mlp.shared_expert.up_proj.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.mlp.shared_expert.gate_proj.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.mlp.shared_expert.down_proj.weight": "model-00028-of-00049.safetensors",
        "model.layers.28.mlp.experts.gate_up_proj": "model-00029-of-00049.safetensors",
        "model.layers.28.mlp.experts.down_proj": "model-00029-of-00049.safetensors",
        "model.layers.28.attn_res_proj.weight": "model-00029-of-00049.safetensors",
        "model.layers.28.attn_res_norm_weight": "model-00029-of-00049.safetensors",
        "model.layers.28.mlp_res_proj.weight": "model-00029-of-00049.safetensors",
        "model.layers.28.mlp_res_norm_weight": "model-00029-of-00049.safetensors",
        "model.layers.29.input_layernorm.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.post_attention_layernorm.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.linear_attn.q_conv1d.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.linear_attn.k_conv1d.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.linear_attn.v_conv1d.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.linear_attn.q_proj.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.linear_attn.k_proj.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.linear_attn.v_proj.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.linear_attn.f_a_proj.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.linear_attn.f_b_proj.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.linear_attn.b_proj.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.linear_attn.g_a_proj.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.linear_attn.g_b_proj.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.linear_attn.o_proj.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.linear_attn.a_log_bias": "model-00029-of-00049.safetensors",
        "model.layers.29.linear_attn.dt_bias": "model-00029-of-00049.safetensors",
        "model.layers.29.linear_attn.o_norm.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.mlp.gate.e_score_correction_bias": "model-00029-of-00049.safetensors",
        "model.layers.29.mlp.gate.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.mlp.shared_expert_gate.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.mlp.shared_expert.up_proj.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.mlp.shared_expert.gate_proj.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.mlp.shared_expert.down_proj.weight": "model-00029-of-00049.safetensors",
        "model.layers.29.mlp.experts.gate_up_proj": "model-00030-of-00049.safetensors",
        "model.layers.29.mlp.experts.down_proj": "model-00030-of-00049.safetensors",
        "model.layers.29.attn_res_proj.weight": "model-00030-of-00049.safetensors",
        "model.layers.29.attn_res_norm_weight": "model-00030-of-00049.safetensors",
        "model.layers.29.mlp_res_proj.weight": "model-00030-of-00049.safetensors",
        "model.layers.29.mlp_res_norm_weight": "model-00030-of-00049.safetensors",
        "model.layers.30.input_layernorm.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.post_attention_layernorm.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.linear_attn.q_conv1d.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.linear_attn.k_conv1d.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.linear_attn.v_conv1d.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.linear_attn.q_proj.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.linear_attn.k_proj.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.linear_attn.v_proj.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.linear_attn.f_a_proj.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.linear_attn.f_b_proj.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.linear_attn.b_proj.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.linear_attn.g_a_proj.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.linear_attn.g_b_proj.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.linear_attn.o_proj.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.linear_attn.a_log_bias": "model-00030-of-00049.safetensors",
        "model.layers.30.linear_attn.dt_bias": "model-00030-of-00049.safetensors",
        "model.layers.30.linear_attn.o_norm.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.mlp.gate.e_score_correction_bias": "model-00030-of-00049.safetensors",
        "model.layers.30.mlp.gate.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.mlp.shared_expert_gate.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.mlp.shared_expert.up_proj.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.mlp.shared_expert.gate_proj.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.mlp.shared_expert.down_proj.weight": "model-00030-of-00049.safetensors",
        "model.layers.30.mlp.experts.gate_up_proj": "model-00031-of-00049.safetensors",
        "model.layers.30.mlp.experts.down_proj": "model-00031-of-00049.safetensors",
        "model.layers.30.attn_res_proj.weight": "model-00031-of-00049.safetensors",
        "model.layers.30.attn_res_norm_weight": "model-00031-of-00049.safetensors",
        "model.layers.30.mlp_res_proj.weight": "model-00031-of-00049.safetensors",
        "model.layers.30.mlp_res_norm_weight": "model-00031-of-00049.safetensors",
        "model.layers.31.input_layernorm.weight": "model-00031-of-00049.safetensors",
        "model.layers.31.post_attention_layernorm.weight": "model-00031-of-00049.safetensors",
        "model.layers.31.self_attn.q_proj.weight": "model-00031-of-00049.safetensors",
        "model.layers.31.self_attn.k_proj.weight": "model-00031-of-00049.safetensors",
        "model.layers.31.self_attn.v_proj.weight": "model-00031-of-00049.safetensors",
        "model.layers.31.self_attn.o_proj.weight": "model-00031-of-00049.safetensors",
        "model.layers.31.self_attn.k_norm.weight": "model-00031-of-00049.safetensors",
        "model.layers.31.self_attn.q_norm.weight": "model-00031-of-00049.safetensors",
        "model.layers.31.mlp.gate.e_score_correction_bias": "model-00031-of-00049.safetensors",
        "model.layers.31.mlp.gate.weight": "model-00031-of-00049.safetensors",
        "model.layers.31.mlp.shared_expert_gate.weight": "model-00031-of-00049.safetensors",
        "model.layers.31.mlp.shared_expert.up_proj.weight": "model-00031-of-00049.safetensors",
        "model.layers.31.mlp.shared_expert.gate_proj.weight": "model-00031-of-00049.safetensors",
        "model.layers.31.mlp.shared_expert.down_proj.weight": "model-00031-of-00049.safetensors",
        "model.layers.31.mlp.experts.gate_up_proj": "model-00032-of-00049.safetensors",
        "model.layers.31.mlp.experts.down_proj": "model-00032-of-00049.safetensors",
        "model.layers.31.attn_res_proj.weight": "model-00032-of-00049.safetensors",
        "model.layers.31.attn_res_norm_weight": "model-00032-of-00049.safetensors",
        "model.layers.31.mlp_res_proj.weight": "model-00032-of-00049.safetensors",
        "model.layers.31.mlp_res_norm_weight": "model-00032-of-00049.safetensors",
        "model.layers.32.input_layernorm.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.post_attention_layernorm.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.linear_attn.q_conv1d.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.linear_attn.k_conv1d.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.linear_attn.v_conv1d.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.linear_attn.q_proj.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.linear_attn.k_proj.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.linear_attn.v_proj.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.linear_attn.f_a_proj.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.linear_attn.f_b_proj.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.linear_attn.b_proj.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.linear_attn.g_a_proj.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.linear_attn.g_b_proj.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.linear_attn.o_proj.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.linear_attn.a_log_bias": "model-00032-of-00049.safetensors",
        "model.layers.32.linear_attn.dt_bias": "model-00032-of-00049.safetensors",
        "model.layers.32.linear_attn.o_norm.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.mlp.gate.e_score_correction_bias": "model-00032-of-00049.safetensors",
        "model.layers.32.mlp.gate.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.mlp.shared_expert_gate.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.mlp.shared_expert.up_proj.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.mlp.shared_expert.gate_proj.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.mlp.shared_expert.down_proj.weight": "model-00032-of-00049.safetensors",
        "model.layers.32.mlp.experts.gate_up_proj": "model-00033-of-00049.safetensors",
        "model.layers.32.mlp.experts.down_proj": "model-00033-of-00049.safetensors",
        "model.layers.32.attn_res_proj.weight": "model-00033-of-00049.safetensors",
        "model.layers.32.attn_res_norm_weight": "model-00033-of-00049.safetensors",
        "model.layers.32.mlp_res_proj.weight": "model-00033-of-00049.safetensors",
        "model.layers.32.mlp_res_norm_weight": "model-00033-of-00049.safetensors",
        "model.layers.33.input_layernorm.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.post_attention_layernorm.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.linear_attn.q_conv1d.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.linear_attn.k_conv1d.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.linear_attn.v_conv1d.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.linear_attn.q_proj.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.linear_attn.k_proj.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.linear_attn.v_proj.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.linear_attn.f_a_proj.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.linear_attn.f_b_proj.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.linear_attn.b_proj.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.linear_attn.g_a_proj.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.linear_attn.g_b_proj.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.linear_attn.o_proj.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.linear_attn.a_log_bias": "model-00033-of-00049.safetensors",
        "model.layers.33.linear_attn.dt_bias": "model-00033-of-00049.safetensors",
        "model.layers.33.linear_attn.o_norm.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.mlp.gate.e_score_correction_bias": "model-00033-of-00049.safetensors",
        "model.layers.33.mlp.gate.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.mlp.shared_expert_gate.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.mlp.shared_expert.up_proj.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.mlp.shared_expert.gate_proj.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.mlp.shared_expert.down_proj.weight": "model-00033-of-00049.safetensors",
        "model.layers.33.mlp.experts.gate_up_proj": "model-00034-of-00049.safetensors",
        "model.layers.33.mlp.experts.down_proj": "model-00034-of-00049.safetensors",
        "model.layers.33.attn_res_proj.weight": "model-00034-of-00049.safetensors",
        "model.layers.33.attn_res_norm_weight": "model-00034-of-00049.safetensors",
        "model.layers.33.mlp_res_proj.weight": "model-00034-of-00049.safetensors",
        "model.layers.33.mlp_res_norm_weight": "model-00034-of-00049.safetensors",
        "model.layers.34.input_layernorm.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.post_attention_layernorm.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.linear_attn.q_conv1d.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.linear_attn.k_conv1d.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.linear_attn.v_conv1d.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.linear_attn.q_proj.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.linear_attn.k_proj.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.linear_attn.v_proj.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.linear_attn.f_a_proj.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.linear_attn.f_b_proj.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.linear_attn.b_proj.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.linear_attn.g_a_proj.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.linear_attn.g_b_proj.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.linear_attn.o_proj.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.linear_attn.a_log_bias": "model-00034-of-00049.safetensors",
        "model.layers.34.linear_attn.dt_bias": "model-00034-of-00049.safetensors",
        "model.layers.34.linear_attn.o_norm.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.mlp.gate.e_score_correction_bias": "model-00034-of-00049.safetensors",
        "model.layers.34.mlp.gate.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.mlp.shared_expert_gate.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.mlp.shared_expert.up_proj.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.mlp.shared_expert.gate_proj.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.mlp.shared_expert.down_proj.weight": "model-00034-of-00049.safetensors",
        "model.layers.34.mlp.experts.gate_up_proj": "model-00035-of-00049.safetensors",
        "model.layers.34.mlp.experts.down_proj": "model-00035-of-00049.safetensors",
        "model.layers.34.attn_res_proj.weight": "model-00035-of-00049.safetensors",
        "model.layers.34.attn_res_norm_weight": "model-00035-of-00049.safetensors",
        "model.layers.34.mlp_res_proj.weight": "model-00035-of-00049.safetensors",
        "model.layers.34.mlp_res_norm_weight": "model-00035-of-00049.safetensors",
        "model.layers.35.input_layernorm.weight": "model-00035-of-00049.safetensors",
        "model.layers.35.post_attention_layernorm.weight": "model-00035-of-00049.safetensors",
        "model.layers.35.self_attn.q_proj.weight": "model-00035-of-00049.safetensors",
        "model.layers.35.self_attn.k_proj.weight": "model-00035-of-00049.safetensors",
        "model.layers.35.self_attn.v_proj.weight": "model-00035-of-00049.safetensors",
        "model.layers.35.self_attn.o_proj.weight": "model-00035-of-00049.safetensors",
        "model.layers.35.self_attn.k_norm.weight": "model-00035-of-00049.safetensors",
        "model.layers.35.self_attn.q_norm.weight": "model-00035-of-00049.safetensors",
        "model.layers.35.mlp.gate.e_score_correction_bias": "model-00035-of-00049.safetensors",
        "model.layers.35.mlp.gate.weight": "model-00035-of-00049.safetensors",
        "model.layers.35.mlp.shared_expert_gate.weight": "model-00035-of-00049.safetensors",
        "model.layers.35.mlp.shared_expert.up_proj.weight": "model-00035-of-00049.safetensors",
        "model.layers.35.mlp.shared_expert.gate_proj.weight": "model-00035-of-00049.safetensors",
        "model.layers.35.mlp.shared_expert.down_proj.weight": "model-00035-of-00049.safetensors",
        "model.layers.35.mlp.experts.gate_up_proj": "model-00036-of-00049.safetensors",
        "model.layers.35.mlp.experts.down_proj": "model-00036-of-00049.safetensors",
        "model.layers.35.attn_res_proj.weight": "model-00036-of-00049.safetensors",
        "model.layers.35.attn_res_norm_weight": "model-00036-of-00049.safetensors",
        "model.layers.35.mlp_res_proj.weight": "model-00036-of-00049.safetensors",
        "model.layers.35.mlp_res_norm_weight": "model-00036-of-00049.safetensors",
        "model.layers.36.input_layernorm.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.post_attention_layernorm.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.linear_attn.q_conv1d.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.linear_attn.k_conv1d.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.linear_attn.v_conv1d.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.linear_attn.q_proj.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.linear_attn.k_proj.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.linear_attn.v_proj.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.linear_attn.f_a_proj.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.linear_attn.f_b_proj.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.linear_attn.b_proj.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.linear_attn.g_a_proj.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.linear_attn.g_b_proj.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.linear_attn.o_proj.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.linear_attn.a_log_bias": "model-00036-of-00049.safetensors",
        "model.layers.36.linear_attn.dt_bias": "model-00036-of-00049.safetensors",
        "model.layers.36.linear_attn.o_norm.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.mlp.gate.e_score_correction_bias": "model-00036-of-00049.safetensors",
        "model.layers.36.mlp.gate.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.mlp.shared_expert_gate.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.mlp.shared_expert.up_proj.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.mlp.shared_expert.gate_proj.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.mlp.shared_expert.down_proj.weight": "model-00036-of-00049.safetensors",
        "model.layers.36.mlp.experts.gate_up_proj": "model-00037-of-00049.safetensors",
        "model.layers.36.mlp.experts.down_proj": "model-00037-of-00049.safetensors",
        "model.layers.36.attn_res_proj.weight": "model-00037-of-00049.safetensors",
        "model.layers.36.attn_res_norm_weight": "model-00037-of-00049.safetensors",
        "model.layers.36.mlp_res_proj.weight": "model-00037-of-00049.safetensors",
        "model.layers.36.mlp_res_norm_weight": "model-00037-of-00049.safetensors",
        "model.layers.37.input_layernorm.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.post_attention_layernorm.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.linear_attn.q_conv1d.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.linear_attn.k_conv1d.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.linear_attn.v_conv1d.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.linear_attn.q_proj.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.linear_attn.k_proj.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.linear_attn.v_proj.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.linear_attn.f_a_proj.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.linear_attn.f_b_proj.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.linear_attn.b_proj.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.linear_attn.g_a_proj.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.linear_attn.g_b_proj.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.linear_attn.o_proj.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.linear_attn.a_log_bias": "model-00037-of-00049.safetensors",
        "model.layers.37.linear_attn.dt_bias": "model-00037-of-00049.safetensors",
        "model.layers.37.linear_attn.o_norm.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.mlp.gate.e_score_correction_bias": "model-00037-of-00049.safetensors",
        "model.layers.37.mlp.gate.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.mlp.shared_expert_gate.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.mlp.shared_expert.up_proj.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.mlp.shared_expert.gate_proj.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.mlp.shared_expert.down_proj.weight": "model-00037-of-00049.safetensors",
        "model.layers.37.mlp.experts.gate_up_proj": "model-00038-of-00049.safetensors",
        "model.layers.37.mlp.experts.down_proj": "model-00038-of-00049.safetensors",
        "model.layers.37.attn_res_proj.weight": "model-00038-of-00049.safetensors",
        "model.layers.37.attn_res_norm_weight": "model-00038-of-00049.safetensors",
        "model.layers.37.mlp_res_proj.weight": "model-00038-of-00049.safetensors",
        "model.layers.37.mlp_res_norm_weight": "model-00038-of-00049.safetensors",
        "model.layers.38.input_layernorm.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.post_attention_layernorm.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.linear_attn.q_conv1d.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.linear_attn.k_conv1d.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.linear_attn.v_conv1d.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.linear_attn.q_proj.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.linear_attn.k_proj.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.linear_attn.v_proj.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.linear_attn.f_a_proj.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.linear_attn.f_b_proj.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.linear_attn.b_proj.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.linear_attn.g_a_proj.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.linear_attn.g_b_proj.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.linear_attn.o_proj.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.linear_attn.a_log_bias": "model-00038-of-00049.safetensors",
        "model.layers.38.linear_attn.dt_bias": "model-00038-of-00049.safetensors",
        "model.layers.38.linear_attn.o_norm.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.mlp.gate.e_score_correction_bias": "model-00038-of-00049.safetensors",
        "model.layers.38.mlp.gate.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.mlp.shared_expert_gate.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.mlp.shared_expert.up_proj.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.mlp.shared_expert.gate_proj.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.mlp.shared_expert.down_proj.weight": "model-00038-of-00049.safetensors",
        "model.layers.38.mlp.experts.gate_up_proj": "model-00039-of-00049.safetensors",
        "model.layers.38.mlp.experts.down_proj": "model-00039-of-00049.safetensors",
        "model.layers.38.attn_res_proj.weight": "model-00039-of-00049.safetensors",
        "model.layers.38.attn_res_norm_weight": "model-00039-of-00049.safetensors",
        "model.layers.38.mlp_res_proj.weight": "model-00039-of-00049.safetensors",
        "model.layers.38.mlp_res_norm_weight": "model-00039-of-00049.safetensors",
        "model.layers.39.input_layernorm.weight": "model-00039-of-00049.safetensors",
        "model.layers.39.post_attention_layernorm.weight": "model-00039-of-00049.safetensors",
        "model.layers.39.self_attn.q_proj.weight": "model-00039-of-00049.safetensors",
        "model.layers.39.self_attn.k_proj.weight": "model-00039-of-00049.safetensors",
        "model.layers.39.self_attn.v_proj.weight": "model-00039-of-00049.safetensors",
        "model.layers.39.self_attn.o_proj.weight": "model-00039-of-00049.safetensors",
        "model.layers.39.self_attn.k_norm.weight": "model-00039-of-00049.safetensors",
        "model.layers.39.self_attn.q_norm.weight": "model-00039-of-00049.safetensors",
        "model.layers.39.mlp.gate.e_score_correction_bias": "model-00039-of-00049.safetensors",
        "model.layers.39.mlp.gate.weight": "model-00039-of-00049.safetensors",
        "model.layers.39.mlp.shared_expert_gate.weight": "model-00039-of-00049.safetensors",
        "model.layers.39.mlp.shared_expert.up_proj.weight": "model-00039-of-00049.safetensors",
        "model.layers.39.mlp.shared_expert.gate_proj.weight": "model-00039-of-00049.safetensors",
        "model.layers.39.mlp.shared_expert.down_proj.weight": "model-00039-of-00049.safetensors",
        "model.layers.39.mlp.experts.gate_up_proj": "model-00040-of-00049.safetensors",
        "model.layers.39.mlp.experts.down_proj": "model-00040-of-00049.safetensors",
        "model.layers.39.attn_res_proj.weight": "model-00040-of-00049.safetensors",
        "model.layers.39.attn_res_norm_weight": "model-00040-of-00049.safetensors",
        "model.layers.39.mlp_res_proj.weight": "model-00040-of-00049.safetensors",
        "model.layers.39.mlp_res_norm_weight": "model-00040-of-00049.safetensors",
        "model.layers.40.input_layernorm.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.post_attention_layernorm.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.linear_attn.q_conv1d.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.linear_attn.k_conv1d.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.linear_attn.v_conv1d.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.linear_attn.q_proj.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.linear_attn.k_proj.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.linear_attn.v_proj.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.linear_attn.f_a_proj.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.linear_attn.f_b_proj.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.linear_attn.b_proj.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.linear_attn.g_a_proj.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.linear_attn.g_b_proj.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.linear_attn.o_proj.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.linear_attn.a_log_bias": "model-00040-of-00049.safetensors",
        "model.layers.40.linear_attn.dt_bias": "model-00040-of-00049.safetensors",
        "model.layers.40.linear_attn.o_norm.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.mlp.gate.e_score_correction_bias": "model-00040-of-00049.safetensors",
        "model.layers.40.mlp.gate.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.mlp.shared_expert_gate.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.mlp.shared_expert.up_proj.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.mlp.shared_expert.gate_proj.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.mlp.shared_expert.down_proj.weight": "model-00040-of-00049.safetensors",
        "model.layers.40.mlp.experts.gate_up_proj": "model-00041-of-00049.safetensors",
        "model.layers.40.mlp.experts.down_proj": "model-00041-of-00049.safetensors",
        "model.layers.40.attn_res_proj.weight": "model-00041-of-00049.safetensors",
        "model.layers.40.attn_res_norm_weight": "model-00041-of-00049.safetensors",
        "model.layers.40.mlp_res_proj.weight": "model-00041-of-00049.safetensors",
        "model.layers.40.mlp_res_norm_weight": "model-00041-of-00049.safetensors",
        "model.layers.41.input_layernorm.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.post_attention_layernorm.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.linear_attn.q_conv1d.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.linear_attn.k_conv1d.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.linear_attn.v_conv1d.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.linear_attn.q_proj.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.linear_attn.k_proj.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.linear_attn.v_proj.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.linear_attn.f_a_proj.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.linear_attn.f_b_proj.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.linear_attn.b_proj.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.linear_attn.g_a_proj.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.linear_attn.g_b_proj.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.linear_attn.o_proj.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.linear_attn.a_log_bias": "model-00041-of-00049.safetensors",
        "model.layers.41.linear_attn.dt_bias": "model-00041-of-00049.safetensors",
        "model.layers.41.linear_attn.o_norm.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.mlp.gate.e_score_correction_bias": "model-00041-of-00049.safetensors",
        "model.layers.41.mlp.gate.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.mlp.shared_expert_gate.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.mlp.shared_expert.up_proj.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.mlp.shared_expert.gate_proj.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.mlp.shared_expert.down_proj.weight": "model-00041-of-00049.safetensors",
        "model.layers.41.mlp.experts.gate_up_proj": "model-00042-of-00049.safetensors",
        "model.layers.41.mlp.experts.down_proj": "model-00042-of-00049.safetensors",
        "model.layers.41.attn_res_proj.weight": "model-00042-of-00049.safetensors",
        "model.layers.41.attn_res_norm_weight": "model-00042-of-00049.safetensors",
        "model.layers.41.mlp_res_proj.weight": "model-00042-of-00049.safetensors",
        "model.layers.41.mlp_res_norm_weight": "model-00042-of-00049.safetensors",
        "model.layers.42.input_layernorm.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.post_attention_layernorm.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.linear_attn.q_conv1d.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.linear_attn.k_conv1d.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.linear_attn.v_conv1d.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.linear_attn.q_proj.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.linear_attn.k_proj.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.linear_attn.v_proj.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.linear_attn.f_a_proj.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.linear_attn.f_b_proj.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.linear_attn.b_proj.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.linear_attn.g_a_proj.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.linear_attn.g_b_proj.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.linear_attn.o_proj.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.linear_attn.a_log_bias": "model-00042-of-00049.safetensors",
        "model.layers.42.linear_attn.dt_bias": "model-00042-of-00049.safetensors",
        "model.layers.42.linear_attn.o_norm.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.mlp.gate.e_score_correction_bias": "model-00042-of-00049.safetensors",
        "model.layers.42.mlp.gate.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.mlp.shared_expert_gate.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.mlp.shared_expert.up_proj.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.mlp.shared_expert.gate_proj.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.mlp.shared_expert.down_proj.weight": "model-00042-of-00049.safetensors",
        "model.layers.42.mlp.experts.gate_up_proj": "model-00043-of-00049.safetensors",
        "model.layers.42.mlp.experts.down_proj": "model-00043-of-00049.safetensors",
        "model.layers.42.attn_res_proj.weight": "model-00043-of-00049.safetensors",
        "model.layers.42.attn_res_norm_weight": "model-00043-of-00049.safetensors",
        "model.layers.42.mlp_res_proj.weight": "model-00043-of-00049.safetensors",
        "model.layers.42.mlp_res_norm_weight": "model-00043-of-00049.safetensors",
        "model.layers.43.input_layernorm.weight": "model-00043-of-00049.safetensors",
        "model.layers.43.post_attention_layernorm.weight": "model-00043-of-00049.safetensors",
        "model.layers.43.self_attn.q_proj.weight": "model-00043-of-00049.safetensors",
        "model.layers.43.self_attn.k_proj.weight": "model-00043-of-00049.safetensors",
        "model.layers.43.self_attn.v_proj.weight": "model-00043-of-00049.safetensors",
        "model.layers.43.self_attn.o_proj.weight": "model-00043-of-00049.safetensors",
        "model.layers.43.self_attn.k_norm.weight": "model-00043-of-00049.safetensors",
        "model.layers.43.self_attn.q_norm.weight": "model-00043-of-00049.safetensors",
        "model.layers.43.mlp.gate.e_score_correction_bias": "model-00043-of-00049.safetensors",
        "model.layers.43.mlp.gate.weight": "model-00043-of-00049.safetensors",
        "model.layers.43.mlp.shared_expert_gate.weight": "model-00043-of-00049.safetensors",
        "model.layers.43.mlp.shared_expert.up_proj.weight": "model-00043-of-00049.safetensors",
        "model.layers.43.mlp.shared_expert.gate_proj.weight": "model-00043-of-00049.safetensors",
        "model.layers.43.mlp.shared_expert.down_proj.weight": "model-00043-of-00049.safetensors",
        "model.layers.43.mlp.experts.gate_up_proj": "model-00044-of-00049.safetensors",
        "model.layers.43.mlp.experts.down_proj": "model-00044-of-00049.safetensors",
        "model.layers.43.attn_res_proj.weight": "model-00044-of-00049.safetensors",
        "model.layers.43.attn_res_norm_weight": "model-00044-of-00049.safetensors",
        "model.layers.43.mlp_res_proj.weight": "model-00044-of-00049.safetensors",
        "model.layers.43.mlp_res_norm_weight": "model-00044-of-00049.safetensors",
        "model.layers.44.input_layernorm.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.post_attention_layernorm.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.linear_attn.q_conv1d.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.linear_attn.k_conv1d.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.linear_attn.v_conv1d.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.linear_attn.q_proj.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.linear_attn.k_proj.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.linear_attn.v_proj.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.linear_attn.f_a_proj.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.linear_attn.f_b_proj.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.linear_attn.b_proj.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.linear_attn.g_a_proj.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.linear_attn.g_b_proj.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.linear_attn.o_proj.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.linear_attn.a_log_bias": "model-00044-of-00049.safetensors",
        "model.layers.44.linear_attn.dt_bias": "model-00044-of-00049.safetensors",
        "model.layers.44.linear_attn.o_norm.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.mlp.gate.e_score_correction_bias": "model-00044-of-00049.safetensors",
        "model.layers.44.mlp.gate.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.mlp.shared_expert_gate.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.mlp.shared_expert.up_proj.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.mlp.shared_expert.gate_proj.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.mlp.shared_expert.down_proj.weight": "model-00044-of-00049.safetensors",
        "model.layers.44.mlp.experts.gate_up_proj": "model-00045-of-00049.safetensors",
        "model.layers.44.mlp.experts.down_proj": "model-00045-of-00049.safetensors",
        "model.layers.44.attn_res_proj.weight": "model-00045-of-00049.safetensors",
        "model.layers.44.attn_res_norm_weight": "model-00045-of-00049.safetensors",
        "model.layers.44.mlp_res_proj.weight": "model-00045-of-00049.safetensors",
        "model.layers.44.mlp_res_norm_weight": "model-00045-of-00049.safetensors",
        "model.layers.45.input_layernorm.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.post_attention_layernorm.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.linear_attn.q_conv1d.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.linear_attn.k_conv1d.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.linear_attn.v_conv1d.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.linear_attn.q_proj.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.linear_attn.k_proj.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.linear_attn.v_proj.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.linear_attn.f_a_proj.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.linear_attn.f_b_proj.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.linear_attn.b_proj.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.linear_attn.g_a_proj.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.linear_attn.g_b_proj.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.linear_attn.o_proj.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.linear_attn.a_log_bias": "model-00045-of-00049.safetensors",
        "model.layers.45.linear_attn.dt_bias": "model-00045-of-00049.safetensors",
        "model.layers.45.linear_attn.o_norm.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.mlp.gate.e_score_correction_bias": "model-00045-of-00049.safetensors",
        "model.layers.45.mlp.gate.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.mlp.shared_expert_gate.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.mlp.shared_expert.up_proj.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.mlp.shared_expert.gate_proj.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.mlp.shared_expert.down_proj.weight": "model-00045-of-00049.safetensors",
        "model.layers.45.mlp.experts.gate_up_proj": "model-00046-of-00049.safetensors",
        "model.layers.45.mlp.experts.down_proj": "model-00046-of-00049.safetensors",
        "model.layers.45.attn_res_proj.weight": "model-00046-of-00049.safetensors",
        "model.layers.45.attn_res_norm_weight": "model-00046-of-00049.safetensors",
        "model.layers.45.mlp_res_proj.weight": "model-00046-of-00049.safetensors",
        "model.layers.45.mlp_res_norm_weight": "model-00046-of-00049.safetensors",
        "model.layers.46.input_layernorm.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.post_attention_layernorm.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.linear_attn.q_conv1d.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.linear_attn.k_conv1d.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.linear_attn.v_conv1d.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.linear_attn.q_proj.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.linear_attn.k_proj.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.linear_attn.v_proj.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.linear_attn.f_a_proj.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.linear_attn.f_b_proj.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.linear_attn.b_proj.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.linear_attn.g_a_proj.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.linear_attn.g_b_proj.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.linear_attn.o_proj.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.linear_attn.a_log_bias": "model-00046-of-00049.safetensors",
        "model.layers.46.linear_attn.dt_bias": "model-00046-of-00049.safetensors",
        "model.layers.46.linear_attn.o_norm.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.mlp.gate.e_score_correction_bias": "model-00046-of-00049.safetensors",
        "model.layers.46.mlp.gate.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.mlp.shared_expert_gate.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.mlp.shared_expert.up_proj.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.mlp.shared_expert.gate_proj.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.mlp.shared_expert.down_proj.weight": "model-00046-of-00049.safetensors",
        "model.layers.46.mlp.experts.gate_up_proj": "model-00047-of-00049.safetensors",
        "model.layers.46.mlp.experts.down_proj": "model-00047-of-00049.safetensors",
        "model.layers.46.attn_res_proj.weight": "model-00047-of-00049.safetensors",
        "model.layers.46.attn_res_norm_weight": "model-00047-of-00049.safetensors",
        "model.layers.46.mlp_res_proj.weight": "model-00047-of-00049.safetensors",
        "model.layers.46.mlp_res_norm_weight": "model-00047-of-00049.safetensors",
        "model.layers.47.input_layernorm.weight": "model-00047-of-00049.safetensors",
        "model.layers.47.post_attention_layernorm.weight": "model-00047-of-00049.safetensors",
        "model.layers.47.self_attn.q_proj.weight": "model-00047-of-00049.safetensors",
        "model.layers.47.self_attn.k_proj.weight": "model-00047-of-00049.safetensors",
        "model.layers.47.self_attn.v_proj.weight": "model-00047-of-00049.safetensors",
        "model.layers.47.self_attn.o_proj.weight": "model-00047-of-00049.safetensors",
        "model.layers.47.self_attn.k_norm.weight": "model-00047-of-00049.safetensors",
        "model.layers.47.self_attn.q_norm.weight": "model-00047-of-00049.safetensors",
        "model.layers.47.mlp.gate.e_score_correction_bias": "model-00047-of-00049.safetensors",
        "model.layers.47.mlp.gate.weight": "model-00047-of-00049.safetensors",
        "model.layers.47.mlp.shared_expert_gate.weight": "model-00047-of-00049.safetensors",
        "model.layers.47.mlp.shared_expert.up_proj.weight": "model-00047-of-00049.safetensors",
        "model.layers.47.mlp.shared_expert.gate_proj.weight": "model-00047-of-00049.safetensors",
        "model.layers.47.mlp.shared_expert.down_proj.weight": "model-00047-of-00049.safetensors",
        "model.layers.47.mlp.experts.gate_up_proj": "model-00048-of-00049.safetensors",
        "model.layers.47.mlp.experts.down_proj": "model-00048-of-00049.safetensors",
        "model.layers.47.attn_res_proj.weight": "model-00048-of-00049.safetensors",
        "model.layers.47.attn_res_norm_weight": "model-00048-of-00049.safetensors",
        "model.layers.47.mlp_res_proj.weight": "model-00048-of-00049.safetensors",
        "model.layers.47.mlp_res_norm_weight": "model-00048-of-00049.safetensors",
        "model.norm.weight": "model-00048-of-00049.safetensors",
        "model.attnres_final.res_proj.weight": "model-00048-of-00049.safetensors",
        "model.attnres_final.res_norm_weight": "model-00048-of-00049.safetensors",
        "lm_head.weight": "model-00048-of-00049.safetensors",
        "mtp.layers.0.input_layernorm.weight": "model-00048-of-00049.safetensors",
        "mtp.layers.0.post_attention_layernorm.weight": "model-00048-of-00049.safetensors",
        "mtp.layers.0.self_attn.q_proj.weight": "model-00048-of-00049.safetensors",
        "mtp.layers.0.self_attn.k_proj.weight": "model-00048-of-00049.safetensors",
        "mtp.layers.0.self_attn.v_proj.weight": "model-00048-of-00049.safetensors",
        "mtp.layers.0.self_attn.o_proj.weight": "model-00048-of-00049.safetensors",
        "mtp.layers.0.self_attn.k_norm.weight": "model-00048-of-00049.safetensors",
        "mtp.layers.0.self_attn.q_norm.weight": "model-00048-of-00049.safetensors",
        "mtp.layers.0.mlp.gate.e_score_correction_bias": "model-00048-of-00049.safetensors",
        "mtp.layers.0.mlp.gate.weight": "model-00048-of-00049.safetensors",
        "mtp.layers.0.mlp.shared_expert_gate.weight": "model-00048-of-00049.safetensors",
        "mtp.layers.0.mlp.shared_expert.up_proj.weight": "model-00048-of-00049.safetensors",
        "mtp.layers.0.mlp.shared_expert.gate_proj.weight": "model-00048-of-00049.safetensors",
        "mtp.layers.0.mlp.shared_expert.down_proj.weight": "model-00048-of-00049.safetensors",
        "mtp.layers.0.mlp.experts.gate_up_proj": "model-00049-of-00049.safetensors",
        "mtp.layers.0.mlp.experts.down_proj": "model-00049-of-00049.safetensors",
        "mtp.fc.weight": "model-00049-of-00049.safetensors",
        "mtp.pre_fc_norm_embedding.weight": "model-00049-of-00049.safetensors",
        "mtp.pre_fc_norm_hidden.weight": "model-00049-of-00049.safetensors",
        "mtp.norm.weight": "model-00049-of-00049.safetensors"
    }
}