{
  "metadata": {
    "total_size": 4892365451008.0
  },
  "weight_map": {
    "lm_head.weight": "model-00212-of-00213.safetensors",
    "model.embed_tokens.weight": "model-00025-of-00213.safetensors",
    "model.layers.0.input_layernorm.weight": "model-00011-of-00213.safetensors",
    "model.layers.0.linear_attn.A_log": "model-00011-of-00213.safetensors",
    "model.layers.0.linear_attn.conv1d.weight": "model-00011-of-00213.safetensors",
    "model.layers.0.linear_attn.dt_bias": "model-00011-of-00213.safetensors",
    "model.layers.0.linear_attn.in_proj_a.weight": "model-00011-of-00213.safetensors",
    "model.layers.0.linear_attn.in_proj_b.weight": "model-00011-of-00213.safetensors",
    "model.layers.0.linear_attn.in_proj_qkv.weight": "model-00011-of-00213.safetensors",
    "model.layers.0.linear_attn.in_proj_z.weight": "model-00011-of-00213.safetensors",
    "model.layers.0.linear_attn.norm.weight": "model-00011-of-00213.safetensors",
    "model.layers.0.linear_attn.out_proj.weight": "model-00011-of-00213.safetensors",
    "model.layers.0.mlp.experts.down_proj": "model-00002-of-00213.safetensors",
    "model.layers.0.mlp.experts.gate_up_proj": "model-00001-of-00213.safetensors",
    "model.layers.0.mlp.gate.weight": "model-00011-of-00213.safetensors",
    "model.layers.0.mlp.shared_expert.down_proj.weight": "model-00011-of-00213.safetensors",
    "model.layers.0.mlp.shared_expert.gate_proj.weight": "model-00011-of-00213.safetensors",
    "model.layers.0.mlp.shared_expert.up_proj.weight": "model-00011-of-00213.safetensors",
    "model.layers.0.mlp.shared_expert_gate.weight": "model-00011-of-00213.safetensors",
    "model.layers.0.post_attention_layernorm.weight": "model-00011-of-00213.safetensors",
    "model.layers.1.input_layernorm.weight": "model-00011-of-00213.safetensors",
    "model.layers.1.linear_attn.A_log": "model-00011-of-00213.safetensors",
    "model.layers.1.linear_attn.conv1d.weight": "model-00011-of-00213.safetensors",
    "model.layers.1.linear_attn.dt_bias": "model-00011-of-00213.safetensors",
    "model.layers.1.linear_attn.in_proj_a.weight": "model-00011-of-00213.safetensors",
    "model.layers.1.linear_attn.in_proj_b.weight": "model-00011-of-00213.safetensors",
    "model.layers.1.linear_attn.in_proj_qkv.weight": "model-00011-of-00213.safetensors",
    "model.layers.1.linear_attn.in_proj_z.weight": "model-00011-of-00213.safetensors",
    "model.layers.1.linear_attn.norm.weight": "model-00011-of-00213.safetensors",
    "model.layers.1.linear_attn.out_proj.weight": "model-00011-of-00213.safetensors",
    "model.layers.1.mlp.experts.down_proj": "model-00004-of-00213.safetensors",
    "model.layers.1.mlp.experts.gate_up_proj": "model-00003-of-00213.safetensors",
    "model.layers.1.mlp.gate.weight": "model-00011-of-00213.safetensors",
    "model.layers.1.mlp.shared_expert.down_proj.weight": "model-00011-of-00213.safetensors",
    "model.layers.1.mlp.shared_expert.gate_proj.weight": "model-00011-of-00213.safetensors",
    "model.layers.1.mlp.shared_expert.up_proj.weight": "model-00011-of-00213.safetensors",
    "model.layers.1.mlp.shared_expert_gate.weight": "model-00011-of-00213.safetensors",
    "model.layers.1.post_attention_layernorm.weight": "model-00011-of-00213.safetensors",
    "model.layers.10.input_layernorm.weight": "model-00011-of-00213.safetensors",
    "model.layers.10.linear_attn.A_log": "model-00011-of-00213.safetensors",
    "model.layers.10.linear_attn.conv1d.weight": "model-00011-of-00213.safetensors",
    "model.layers.10.linear_attn.dt_bias": "model-00011-of-00213.safetensors",
    "model.layers.10.linear_attn.in_proj_a.weight": "model-00011-of-00213.safetensors",
    "model.layers.10.linear_attn.in_proj_b.weight": "model-00011-of-00213.safetensors",
    "model.layers.10.linear_attn.in_proj_qkv.weight": "model-00011-of-00213.safetensors",
    "model.layers.10.linear_attn.in_proj_z.weight": "model-00011-of-00213.safetensors",
    "model.layers.10.linear_attn.norm.weight": "model-00011-of-00213.safetensors",
    "model.layers.10.linear_attn.out_proj.weight": "model-00011-of-00213.safetensors",
    "model.layers.10.mlp.experts.down_proj": "model-00006-of-00213.safetensors",
    "model.layers.10.mlp.experts.gate_up_proj": "model-00005-of-00213.safetensors",
    "model.layers.10.mlp.gate.weight": "model-00011-of-00213.safetensors",
    "model.layers.10.mlp.shared_expert.down_proj.weight": "model-00011-of-00213.safetensors",
    "model.layers.10.mlp.shared_expert.gate_proj.weight": "model-00011-of-00213.safetensors",
    "model.layers.10.mlp.shared_expert.up_proj.weight": "model-00011-of-00213.safetensors",
    "model.layers.10.mlp.shared_expert_gate.weight": "model-00011-of-00213.safetensors",
    "model.layers.10.post_attention_layernorm.weight": "model-00011-of-00213.safetensors",
    "model.layers.11.input_layernorm.weight": "model-00037-of-00213.safetensors",
    "model.layers.11.mlp.experts.down_proj": "model-00028-of-00213.safetensors",
    "model.layers.11.mlp.experts.gate_up_proj": "model-00027-of-00213.safetensors",
    "model.layers.11.mlp.gate.weight": "model-00037-of-00213.safetensors",
    "model.layers.11.mlp.shared_expert.down_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.11.mlp.shared_expert.gate_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.11.mlp.shared_expert.up_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.11.mlp.shared_expert_gate.weight": "model-00037-of-00213.safetensors",
    "model.layers.11.post_attention_layernorm.weight": "model-00037-of-00213.safetensors",
    "model.layers.11.self_attn.k_norm.weight": "model-00037-of-00213.safetensors",
    "model.layers.11.self_attn.k_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.11.self_attn.o_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.11.self_attn.q_norm.weight": "model-00037-of-00213.safetensors",
    "model.layers.11.self_attn.q_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.11.self_attn.v_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.12.input_layernorm.weight": "model-00037-of-00213.safetensors",
    "model.layers.12.linear_attn.A_log": "model-00037-of-00213.safetensors",
    "model.layers.12.linear_attn.conv1d.weight": "model-00037-of-00213.safetensors",
    "model.layers.12.linear_attn.dt_bias": "model-00037-of-00213.safetensors",
    "model.layers.12.linear_attn.in_proj_a.weight": "model-00037-of-00213.safetensors",
    "model.layers.12.linear_attn.in_proj_b.weight": "model-00037-of-00213.safetensors",
    "model.layers.12.linear_attn.in_proj_qkv.weight": "model-00037-of-00213.safetensors",
    "model.layers.12.linear_attn.in_proj_z.weight": "model-00037-of-00213.safetensors",
    "model.layers.12.linear_attn.norm.weight": "model-00037-of-00213.safetensors",
    "model.layers.12.linear_attn.out_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.12.mlp.experts.down_proj": "model-00030-of-00213.safetensors",
    "model.layers.12.mlp.experts.gate_up_proj": "model-00029-of-00213.safetensors",
    "model.layers.12.mlp.gate.weight": "model-00037-of-00213.safetensors",
    "model.layers.12.mlp.shared_expert.down_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.12.mlp.shared_expert.gate_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.12.mlp.shared_expert.up_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.12.mlp.shared_expert_gate.weight": "model-00037-of-00213.safetensors",
    "model.layers.12.post_attention_layernorm.weight": "model-00037-of-00213.safetensors",
    "model.layers.13.input_layernorm.weight": "model-00037-of-00213.safetensors",
    "model.layers.13.linear_attn.A_log": "model-00037-of-00213.safetensors",
    "model.layers.13.linear_attn.conv1d.weight": "model-00037-of-00213.safetensors",
    "model.layers.13.linear_attn.dt_bias": "model-00037-of-00213.safetensors",
    "model.layers.13.linear_attn.in_proj_a.weight": "model-00037-of-00213.safetensors",
    "model.layers.13.linear_attn.in_proj_b.weight": "model-00037-of-00213.safetensors",
    "model.layers.13.linear_attn.in_proj_qkv.weight": "model-00037-of-00213.safetensors",
    "model.layers.13.linear_attn.in_proj_z.weight": "model-00037-of-00213.safetensors",
    "model.layers.13.linear_attn.norm.weight": "model-00037-of-00213.safetensors",
    "model.layers.13.linear_attn.out_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.13.mlp.experts.down_proj": "model-00032-of-00213.safetensors",
    "model.layers.13.mlp.experts.gate_up_proj": "model-00031-of-00213.safetensors",
    "model.layers.13.mlp.gate.weight": "model-00037-of-00213.safetensors",
    "model.layers.13.mlp.shared_expert.down_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.13.mlp.shared_expert.gate_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.13.mlp.shared_expert.up_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.13.mlp.shared_expert_gate.weight": "model-00037-of-00213.safetensors",
    "model.layers.13.post_attention_layernorm.weight": "model-00037-of-00213.safetensors",
    "model.layers.14.input_layernorm.weight": "model-00037-of-00213.safetensors",
    "model.layers.14.linear_attn.A_log": "model-00037-of-00213.safetensors",
    "model.layers.14.linear_attn.conv1d.weight": "model-00037-of-00213.safetensors",
    "model.layers.14.linear_attn.dt_bias": "model-00037-of-00213.safetensors",
    "model.layers.14.linear_attn.in_proj_a.weight": "model-00037-of-00213.safetensors",
    "model.layers.14.linear_attn.in_proj_b.weight": "model-00037-of-00213.safetensors",
    "model.layers.14.linear_attn.in_proj_qkv.weight": "model-00037-of-00213.safetensors",
    "model.layers.14.linear_attn.in_proj_z.weight": "model-00037-of-00213.safetensors",
    "model.layers.14.linear_attn.norm.weight": "model-00037-of-00213.safetensors",
    "model.layers.14.linear_attn.out_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.14.mlp.experts.down_proj": "model-00034-of-00213.safetensors",
    "model.layers.14.mlp.experts.gate_up_proj": "model-00033-of-00213.safetensors",
    "model.layers.14.mlp.gate.weight": "model-00037-of-00213.safetensors",
    "model.layers.14.mlp.shared_expert.down_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.14.mlp.shared_expert.gate_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.14.mlp.shared_expert.up_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.14.mlp.shared_expert_gate.weight": "model-00037-of-00213.safetensors",
    "model.layers.14.post_attention_layernorm.weight": "model-00037-of-00213.safetensors",
    "model.layers.15.input_layernorm.weight": "model-00046-of-00213.safetensors",
    "model.layers.15.mlp.experts.down_proj": "model-00036-of-00213.safetensors",
    "model.layers.15.mlp.experts.gate_up_proj": "model-00035-of-00213.safetensors",
    "model.layers.15.mlp.gate.weight": "model-00037-of-00213.safetensors",
    "model.layers.15.mlp.shared_expert.down_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.15.mlp.shared_expert.gate_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.15.mlp.shared_expert.up_proj.weight": "model-00037-of-00213.safetensors",
    "model.layers.15.mlp.shared_expert_gate.weight": "model-00037-of-00213.safetensors",
    "model.layers.15.post_attention_layernorm.weight": "model-00046-of-00213.safetensors",
    "model.layers.15.self_attn.k_norm.weight": "model-00046-of-00213.safetensors",
    "model.layers.15.self_attn.k_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.15.self_attn.o_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.15.self_attn.q_norm.weight": "model-00046-of-00213.safetensors",
    "model.layers.15.self_attn.q_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.15.self_attn.v_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.16.input_layernorm.weight": "model-00046-of-00213.safetensors",
    "model.layers.16.linear_attn.A_log": "model-00046-of-00213.safetensors",
    "model.layers.16.linear_attn.conv1d.weight": "model-00046-of-00213.safetensors",
    "model.layers.16.linear_attn.dt_bias": "model-00046-of-00213.safetensors",
    "model.layers.16.linear_attn.in_proj_a.weight": "model-00046-of-00213.safetensors",
    "model.layers.16.linear_attn.in_proj_b.weight": "model-00046-of-00213.safetensors",
    "model.layers.16.linear_attn.in_proj_qkv.weight": "model-00046-of-00213.safetensors",
    "model.layers.16.linear_attn.in_proj_z.weight": "model-00046-of-00213.safetensors",
    "model.layers.16.linear_attn.norm.weight": "model-00046-of-00213.safetensors",
    "model.layers.16.linear_attn.out_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.16.mlp.experts.down_proj": "model-00039-of-00213.safetensors",
    "model.layers.16.mlp.experts.gate_up_proj": "model-00038-of-00213.safetensors",
    "model.layers.16.mlp.gate.weight": "model-00046-of-00213.safetensors",
    "model.layers.16.mlp.shared_expert.down_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.16.mlp.shared_expert.gate_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.16.mlp.shared_expert.up_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.16.mlp.shared_expert_gate.weight": "model-00046-of-00213.safetensors",
    "model.layers.16.post_attention_layernorm.weight": "model-00046-of-00213.safetensors",
    "model.layers.17.input_layernorm.weight": "model-00046-of-00213.safetensors",
    "model.layers.17.linear_attn.A_log": "model-00046-of-00213.safetensors",
    "model.layers.17.linear_attn.conv1d.weight": "model-00046-of-00213.safetensors",
    "model.layers.17.linear_attn.dt_bias": "model-00046-of-00213.safetensors",
    "model.layers.17.linear_attn.in_proj_a.weight": "model-00046-of-00213.safetensors",
    "model.layers.17.linear_attn.in_proj_b.weight": "model-00046-of-00213.safetensors",
    "model.layers.17.linear_attn.in_proj_qkv.weight": "model-00046-of-00213.safetensors",
    "model.layers.17.linear_attn.in_proj_z.weight": "model-00046-of-00213.safetensors",
    "model.layers.17.linear_attn.norm.weight": "model-00046-of-00213.safetensors",
    "model.layers.17.linear_attn.out_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.17.mlp.experts.down_proj": "model-00041-of-00213.safetensors",
    "model.layers.17.mlp.experts.gate_up_proj": "model-00040-of-00213.safetensors",
    "model.layers.17.mlp.gate.weight": "model-00046-of-00213.safetensors",
    "model.layers.17.mlp.shared_expert.down_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.17.mlp.shared_expert.gate_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.17.mlp.shared_expert.up_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.17.mlp.shared_expert_gate.weight": "model-00046-of-00213.safetensors",
    "model.layers.17.post_attention_layernorm.weight": "model-00046-of-00213.safetensors",
    "model.layers.18.input_layernorm.weight": "model-00046-of-00213.safetensors",
    "model.layers.18.linear_attn.A_log": "model-00046-of-00213.safetensors",
    "model.layers.18.linear_attn.conv1d.weight": "model-00046-of-00213.safetensors",
    "model.layers.18.linear_attn.dt_bias": "model-00046-of-00213.safetensors",
    "model.layers.18.linear_attn.in_proj_a.weight": "model-00046-of-00213.safetensors",
    "model.layers.18.linear_attn.in_proj_b.weight": "model-00046-of-00213.safetensors",
    "model.layers.18.linear_attn.in_proj_qkv.weight": "model-00046-of-00213.safetensors",
    "model.layers.18.linear_attn.in_proj_z.weight": "model-00046-of-00213.safetensors",
    "model.layers.18.linear_attn.norm.weight": "model-00046-of-00213.safetensors",
    "model.layers.18.linear_attn.out_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.18.mlp.experts.down_proj": "model-00043-of-00213.safetensors",
    "model.layers.18.mlp.experts.gate_up_proj": "model-00042-of-00213.safetensors",
    "model.layers.18.mlp.gate.weight": "model-00046-of-00213.safetensors",
    "model.layers.18.mlp.shared_expert.down_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.18.mlp.shared_expert.gate_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.18.mlp.shared_expert.up_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.18.mlp.shared_expert_gate.weight": "model-00046-of-00213.safetensors",
    "model.layers.18.post_attention_layernorm.weight": "model-00046-of-00213.safetensors",
    "model.layers.19.input_layernorm.weight": "model-00051-of-00213.safetensors",
    "model.layers.19.mlp.experts.down_proj": "model-00045-of-00213.safetensors",
    "model.layers.19.mlp.experts.gate_up_proj": "model-00044-of-00213.safetensors",
    "model.layers.19.mlp.gate.weight": "model-00046-of-00213.safetensors",
    "model.layers.19.mlp.shared_expert.down_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.19.mlp.shared_expert.gate_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.19.mlp.shared_expert.up_proj.weight": "model-00046-of-00213.safetensors",
    "model.layers.19.mlp.shared_expert_gate.weight": "model-00046-of-00213.safetensors",
    "model.layers.19.post_attention_layernorm.weight": "model-00046-of-00213.safetensors",
    "model.layers.19.self_attn.k_norm.weight": "model-00046-of-00213.safetensors",
    "model.layers.19.self_attn.k_proj.weight": "model-00051-of-00213.safetensors",
    "model.layers.19.self_attn.o_proj.weight": "model-00051-of-00213.safetensors",
    "model.layers.19.self_attn.q_norm.weight": "model-00051-of-00213.safetensors",
    "model.layers.19.self_attn.q_proj.weight": "model-00051-of-00213.safetensors",
    "model.layers.19.self_attn.v_proj.weight": "model-00051-of-00213.safetensors",
    "model.layers.2.input_layernorm.weight": "model-00011-of-00213.safetensors",
    "model.layers.2.linear_attn.A_log": "model-00011-of-00213.safetensors",
    "model.layers.2.linear_attn.conv1d.weight": "model-00011-of-00213.safetensors",
    "model.layers.2.linear_attn.dt_bias": "model-00011-of-00213.safetensors",
    "model.layers.2.linear_attn.in_proj_a.weight": "model-00011-of-00213.safetensors",
    "model.layers.2.linear_attn.in_proj_b.weight": "model-00011-of-00213.safetensors",
    "model.layers.2.linear_attn.in_proj_qkv.weight": "model-00011-of-00213.safetensors",
    "model.layers.2.linear_attn.in_proj_z.weight": "model-00011-of-00213.safetensors",
    "model.layers.2.linear_attn.norm.weight": "model-00011-of-00213.safetensors",
    "model.layers.2.linear_attn.out_proj.weight": "model-00011-of-00213.safetensors",
    "model.layers.2.mlp.experts.down_proj": "model-00008-of-00213.safetensors",
    "model.layers.2.mlp.experts.gate_up_proj": "model-00007-of-00213.safetensors",
    "model.layers.2.mlp.gate.weight": "model-00011-of-00213.safetensors",
    "model.layers.2.mlp.shared_expert.down_proj.weight": "model-00011-of-00213.safetensors",
    "model.layers.2.mlp.shared_expert.gate_proj.weight": "model-00011-of-00213.safetensors",
    "model.layers.2.mlp.shared_expert.up_proj.weight": "model-00011-of-00213.safetensors",
    "model.layers.2.mlp.shared_expert_gate.weight": "model-00011-of-00213.safetensors",
    "model.layers.2.post_attention_layernorm.weight": "model-00011-of-00213.safetensors",
    "model.layers.20.input_layernorm.weight": "model-00051-of-00213.safetensors",
    "model.layers.20.linear_attn.A_log": "model-00051-of-00213.safetensors",
    "model.layers.20.linear_attn.conv1d.weight": "model-00051-of-00213.safetensors",
    "model.layers.20.linear_attn.dt_bias": "model-00051-of-00213.safetensors",
    "model.layers.20.linear_attn.in_proj_a.weight": "model-00051-of-00213.safetensors",
    "model.layers.20.linear_attn.in_proj_b.weight": "model-00051-of-00213.safetensors",
    "model.layers.20.linear_attn.in_proj_qkv.weight": "model-00051-of-00213.safetensors",
    "model.layers.20.linear_attn.in_proj_z.weight": "model-00051-of-00213.safetensors",
    "model.layers.20.linear_attn.norm.weight": "model-00051-of-00213.safetensors",
    "model.layers.20.linear_attn.out_proj.weight": "model-00051-of-00213.safetensors",
    "model.layers.20.mlp.experts.down_proj": "model-00048-of-00213.safetensors",
    "model.layers.20.mlp.experts.gate_up_proj": "model-00047-of-00213.safetensors",
    "model.layers.20.mlp.gate.weight": "model-00051-of-00213.safetensors",
    "model.layers.20.mlp.shared_expert.down_proj.weight": "model-00051-of-00213.safetensors",
    "model.layers.20.mlp.shared_expert.gate_proj.weight": "model-00051-of-00213.safetensors",
    "model.layers.20.mlp.shared_expert.up_proj.weight": "model-00051-of-00213.safetensors",
    "model.layers.20.mlp.shared_expert_gate.weight": "model-00051-of-00213.safetensors",
    "model.layers.20.post_attention_layernorm.weight": "model-00051-of-00213.safetensors",
    "model.layers.21.input_layernorm.weight": "model-00051-of-00213.safetensors",
    "model.layers.21.linear_attn.A_log": "model-00051-of-00213.safetensors",
    "model.layers.21.linear_attn.conv1d.weight": "model-00051-of-00213.safetensors",
    "model.layers.21.linear_attn.dt_bias": "model-00051-of-00213.safetensors",
    "model.layers.21.linear_attn.in_proj_a.weight": "model-00051-of-00213.safetensors",
    "model.layers.21.linear_attn.in_proj_b.weight": "model-00051-of-00213.safetensors",
    "model.layers.21.linear_attn.in_proj_qkv.weight": "model-00051-of-00213.safetensors",
    "model.layers.21.linear_attn.in_proj_z.weight": "model-00051-of-00213.safetensors",
    "model.layers.21.linear_attn.norm.weight": "model-00051-of-00213.safetensors",
    "model.layers.21.linear_attn.out_proj.weight": "model-00051-of-00213.safetensors",
    "model.layers.21.mlp.experts.down_proj": "model-00050-of-00213.safetensors",
    "model.layers.21.mlp.experts.gate_up_proj": "model-00049-of-00213.safetensors",
    "model.layers.21.mlp.gate.weight": "model-00051-of-00213.safetensors",
    "model.layers.21.mlp.shared_expert.down_proj.weight": "model-00051-of-00213.safetensors",
    "model.layers.21.mlp.shared_expert.gate_proj.weight": "model-00051-of-00213.safetensors",
    "model.layers.21.mlp.shared_expert.up_proj.weight": "model-00051-of-00213.safetensors",
    "model.layers.21.mlp.shared_expert_gate.weight": "model-00051-of-00213.safetensors",
    "model.layers.21.post_attention_layernorm.weight": "model-00051-of-00213.safetensors",
    "model.layers.22.input_layernorm.weight": "model-00060-of-00213.safetensors",
    "model.layers.22.linear_attn.A_log": "model-00060-of-00213.safetensors",
    "model.layers.22.linear_attn.conv1d.weight": "model-00060-of-00213.safetensors",
    "model.layers.22.linear_attn.dt_bias": "model-00060-of-00213.safetensors",
    "model.layers.22.linear_attn.in_proj_a.weight": "model-00060-of-00213.safetensors",
    "model.layers.22.linear_attn.in_proj_b.weight": "model-00060-of-00213.safetensors",
    "model.layers.22.linear_attn.in_proj_qkv.weight": "model-00060-of-00213.safetensors",
    "model.layers.22.linear_attn.in_proj_z.weight": "model-00060-of-00213.safetensors",
    "model.layers.22.linear_attn.norm.weight": "model-00060-of-00213.safetensors",
    "model.layers.22.linear_attn.out_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.22.mlp.experts.down_proj": "model-00053-of-00213.safetensors",
    "model.layers.22.mlp.experts.gate_up_proj": "model-00052-of-00213.safetensors",
    "model.layers.22.mlp.gate.weight": "model-00060-of-00213.safetensors",
    "model.layers.22.mlp.shared_expert.down_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.22.mlp.shared_expert.gate_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.22.mlp.shared_expert.up_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.22.mlp.shared_expert_gate.weight": "model-00060-of-00213.safetensors",
    "model.layers.22.post_attention_layernorm.weight": "model-00060-of-00213.safetensors",
    "model.layers.23.input_layernorm.weight": "model-00060-of-00213.safetensors",
    "model.layers.23.mlp.experts.down_proj": "model-00055-of-00213.safetensors",
    "model.layers.23.mlp.experts.gate_up_proj": "model-00054-of-00213.safetensors",
    "model.layers.23.mlp.gate.weight": "model-00060-of-00213.safetensors",
    "model.layers.23.mlp.shared_expert.down_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.23.mlp.shared_expert.gate_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.23.mlp.shared_expert.up_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.23.mlp.shared_expert_gate.weight": "model-00060-of-00213.safetensors",
    "model.layers.23.post_attention_layernorm.weight": "model-00060-of-00213.safetensors",
    "model.layers.23.self_attn.k_norm.weight": "model-00060-of-00213.safetensors",
    "model.layers.23.self_attn.k_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.23.self_attn.o_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.23.self_attn.q_norm.weight": "model-00060-of-00213.safetensors",
    "model.layers.23.self_attn.q_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.23.self_attn.v_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.24.input_layernorm.weight": "model-00060-of-00213.safetensors",
    "model.layers.24.linear_attn.A_log": "model-00060-of-00213.safetensors",
    "model.layers.24.linear_attn.conv1d.weight": "model-00060-of-00213.safetensors",
    "model.layers.24.linear_attn.dt_bias": "model-00060-of-00213.safetensors",
    "model.layers.24.linear_attn.in_proj_a.weight": "model-00060-of-00213.safetensors",
    "model.layers.24.linear_attn.in_proj_b.weight": "model-00060-of-00213.safetensors",
    "model.layers.24.linear_attn.in_proj_qkv.weight": "model-00060-of-00213.safetensors",
    "model.layers.24.linear_attn.in_proj_z.weight": "model-00060-of-00213.safetensors",
    "model.layers.24.linear_attn.norm.weight": "model-00060-of-00213.safetensors",
    "model.layers.24.linear_attn.out_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.24.mlp.experts.down_proj": "model-00057-of-00213.safetensors",
    "model.layers.24.mlp.experts.gate_up_proj": "model-00056-of-00213.safetensors",
    "model.layers.24.mlp.gate.weight": "model-00060-of-00213.safetensors",
    "model.layers.24.mlp.shared_expert.down_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.24.mlp.shared_expert.gate_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.24.mlp.shared_expert.up_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.24.mlp.shared_expert_gate.weight": "model-00060-of-00213.safetensors",
    "model.layers.24.post_attention_layernorm.weight": "model-00060-of-00213.safetensors",
    "model.layers.25.input_layernorm.weight": "model-00060-of-00213.safetensors",
    "model.layers.25.linear_attn.A_log": "model-00060-of-00213.safetensors",
    "model.layers.25.linear_attn.conv1d.weight": "model-00060-of-00213.safetensors",
    "model.layers.25.linear_attn.dt_bias": "model-00060-of-00213.safetensors",
    "model.layers.25.linear_attn.in_proj_a.weight": "model-00060-of-00213.safetensors",
    "model.layers.25.linear_attn.in_proj_b.weight": "model-00060-of-00213.safetensors",
    "model.layers.25.linear_attn.in_proj_qkv.weight": "model-00060-of-00213.safetensors",
    "model.layers.25.linear_attn.in_proj_z.weight": "model-00060-of-00213.safetensors",
    "model.layers.25.linear_attn.norm.weight": "model-00060-of-00213.safetensors",
    "model.layers.25.linear_attn.out_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.25.mlp.experts.down_proj": "model-00059-of-00213.safetensors",
    "model.layers.25.mlp.experts.gate_up_proj": "model-00058-of-00213.safetensors",
    "model.layers.25.mlp.gate.weight": "model-00060-of-00213.safetensors",
    "model.layers.25.mlp.shared_expert.down_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.25.mlp.shared_expert.gate_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.25.mlp.shared_expert.up_proj.weight": "model-00060-of-00213.safetensors",
    "model.layers.25.mlp.shared_expert_gate.weight": "model-00060-of-00213.safetensors",
    "model.layers.25.post_attention_layernorm.weight": "model-00060-of-00213.safetensors",
    "model.layers.26.input_layernorm.weight": "model-00060-of-00213.safetensors",
    "model.layers.26.linear_attn.A_log": "model-00060-of-00213.safetensors",
    "model.layers.26.linear_attn.conv1d.weight": "model-00060-of-00213.safetensors",
    "model.layers.26.linear_attn.dt_bias": "model-00060-of-00213.safetensors",
    "model.layers.26.linear_attn.in_proj_a.weight": "model-00060-of-00213.safetensors",
    "model.layers.26.linear_attn.in_proj_b.weight": "model-00060-of-00213.safetensors",
    "model.layers.26.linear_attn.in_proj_qkv.weight": "model-00069-of-00213.safetensors",
    "model.layers.26.linear_attn.in_proj_z.weight": "model-00069-of-00213.safetensors",
    "model.layers.26.linear_attn.norm.weight": "model-00069-of-00213.safetensors",
    "model.layers.26.linear_attn.out_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.26.mlp.experts.down_proj": "model-00062-of-00213.safetensors",
    "model.layers.26.mlp.experts.gate_up_proj": "model-00061-of-00213.safetensors",
    "model.layers.26.mlp.gate.weight": "model-00069-of-00213.safetensors",
    "model.layers.26.mlp.shared_expert.down_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.26.mlp.shared_expert.gate_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.26.mlp.shared_expert.up_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.26.mlp.shared_expert_gate.weight": "model-00069-of-00213.safetensors",
    "model.layers.26.post_attention_layernorm.weight": "model-00069-of-00213.safetensors",
    "model.layers.27.input_layernorm.weight": "model-00069-of-00213.safetensors",
    "model.layers.27.mlp.experts.down_proj": "model-00064-of-00213.safetensors",
    "model.layers.27.mlp.experts.gate_up_proj": "model-00063-of-00213.safetensors",
    "model.layers.27.mlp.gate.weight": "model-00069-of-00213.safetensors",
    "model.layers.27.mlp.shared_expert.down_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.27.mlp.shared_expert.gate_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.27.mlp.shared_expert.up_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.27.mlp.shared_expert_gate.weight": "model-00069-of-00213.safetensors",
    "model.layers.27.post_attention_layernorm.weight": "model-00069-of-00213.safetensors",
    "model.layers.27.self_attn.k_norm.weight": "model-00069-of-00213.safetensors",
    "model.layers.27.self_attn.k_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.27.self_attn.o_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.27.self_attn.q_norm.weight": "model-00069-of-00213.safetensors",
    "model.layers.27.self_attn.q_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.27.self_attn.v_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.28.input_layernorm.weight": "model-00069-of-00213.safetensors",
    "model.layers.28.linear_attn.A_log": "model-00069-of-00213.safetensors",
    "model.layers.28.linear_attn.conv1d.weight": "model-00069-of-00213.safetensors",
    "model.layers.28.linear_attn.dt_bias": "model-00069-of-00213.safetensors",
    "model.layers.28.linear_attn.in_proj_a.weight": "model-00069-of-00213.safetensors",
    "model.layers.28.linear_attn.in_proj_b.weight": "model-00069-of-00213.safetensors",
    "model.layers.28.linear_attn.in_proj_qkv.weight": "model-00069-of-00213.safetensors",
    "model.layers.28.linear_attn.in_proj_z.weight": "model-00069-of-00213.safetensors",
    "model.layers.28.linear_attn.norm.weight": "model-00069-of-00213.safetensors",
    "model.layers.28.linear_attn.out_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.28.mlp.experts.down_proj": "model-00066-of-00213.safetensors",
    "model.layers.28.mlp.experts.gate_up_proj": "model-00065-of-00213.safetensors",
    "model.layers.28.mlp.gate.weight": "model-00069-of-00213.safetensors",
    "model.layers.28.mlp.shared_expert.down_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.28.mlp.shared_expert.gate_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.28.mlp.shared_expert.up_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.28.mlp.shared_expert_gate.weight": "model-00069-of-00213.safetensors",
    "model.layers.28.post_attention_layernorm.weight": "model-00069-of-00213.safetensors",
    "model.layers.29.input_layernorm.weight": "model-00069-of-00213.safetensors",
    "model.layers.29.linear_attn.A_log": "model-00069-of-00213.safetensors",
    "model.layers.29.linear_attn.conv1d.weight": "model-00069-of-00213.safetensors",
    "model.layers.29.linear_attn.dt_bias": "model-00069-of-00213.safetensors",
    "model.layers.29.linear_attn.in_proj_a.weight": "model-00069-of-00213.safetensors",
    "model.layers.29.linear_attn.in_proj_b.weight": "model-00069-of-00213.safetensors",
    "model.layers.29.linear_attn.in_proj_qkv.weight": "model-00069-of-00213.safetensors",
    "model.layers.29.linear_attn.in_proj_z.weight": "model-00069-of-00213.safetensors",
    "model.layers.29.linear_attn.norm.weight": "model-00069-of-00213.safetensors",
    "model.layers.29.linear_attn.out_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.29.mlp.experts.down_proj": "model-00068-of-00213.safetensors",
    "model.layers.29.mlp.experts.gate_up_proj": "model-00067-of-00213.safetensors",
    "model.layers.29.mlp.gate.weight": "model-00069-of-00213.safetensors",
    "model.layers.29.mlp.shared_expert.down_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.29.mlp.shared_expert.gate_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.29.mlp.shared_expert.up_proj.weight": "model-00069-of-00213.safetensors",
    "model.layers.29.mlp.shared_expert_gate.weight": "model-00069-of-00213.safetensors",
    "model.layers.29.post_attention_layernorm.weight": "model-00069-of-00213.safetensors",
    "model.layers.3.input_layernorm.weight": "model-00020-of-00213.safetensors",
    "model.layers.3.mlp.experts.down_proj": "model-00010-of-00213.safetensors",
    "model.layers.3.mlp.experts.gate_up_proj": "model-00009-of-00213.safetensors",
    "model.layers.3.mlp.gate.weight": "model-00011-of-00213.safetensors",
    "model.layers.3.mlp.shared_expert.down_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.3.mlp.shared_expert.gate_proj.weight": "model-00011-of-00213.safetensors",
    "model.layers.3.mlp.shared_expert.up_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.3.mlp.shared_expert_gate.weight": "model-00011-of-00213.safetensors",
    "model.layers.3.post_attention_layernorm.weight": "model-00020-of-00213.safetensors",
    "model.layers.3.self_attn.k_norm.weight": "model-00020-of-00213.safetensors",
    "model.layers.3.self_attn.k_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.3.self_attn.o_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.3.self_attn.q_norm.weight": "model-00020-of-00213.safetensors",
    "model.layers.3.self_attn.q_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.3.self_attn.v_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.30.input_layernorm.weight": "model-00069-of-00213.safetensors",
    "model.layers.30.linear_attn.A_log": "model-00069-of-00213.safetensors",
    "model.layers.30.linear_attn.conv1d.weight": "model-00069-of-00213.safetensors",
    "model.layers.30.linear_attn.dt_bias": "model-00069-of-00213.safetensors",
    "model.layers.30.linear_attn.in_proj_a.weight": "model-00069-of-00213.safetensors",
    "model.layers.30.linear_attn.in_proj_b.weight": "model-00069-of-00213.safetensors",
    "model.layers.30.linear_attn.in_proj_qkv.weight": "model-00078-of-00213.safetensors",
    "model.layers.30.linear_attn.in_proj_z.weight": "model-00078-of-00213.safetensors",
    "model.layers.30.linear_attn.norm.weight": "model-00078-of-00213.safetensors",
    "model.layers.30.linear_attn.out_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.30.mlp.experts.down_proj": "model-00071-of-00213.safetensors",
    "model.layers.30.mlp.experts.gate_up_proj": "model-00070-of-00213.safetensors",
    "model.layers.30.mlp.gate.weight": "model-00078-of-00213.safetensors",
    "model.layers.30.mlp.shared_expert.down_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.30.mlp.shared_expert.gate_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.30.mlp.shared_expert.up_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.30.mlp.shared_expert_gate.weight": "model-00078-of-00213.safetensors",
    "model.layers.30.post_attention_layernorm.weight": "model-00078-of-00213.safetensors",
    "model.layers.31.input_layernorm.weight": "model-00078-of-00213.safetensors",
    "model.layers.31.mlp.experts.down_proj": "model-00073-of-00213.safetensors",
    "model.layers.31.mlp.experts.gate_up_proj": "model-00072-of-00213.safetensors",
    "model.layers.31.mlp.gate.weight": "model-00078-of-00213.safetensors",
    "model.layers.31.mlp.shared_expert.down_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.31.mlp.shared_expert.gate_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.31.mlp.shared_expert.up_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.31.mlp.shared_expert_gate.weight": "model-00078-of-00213.safetensors",
    "model.layers.31.post_attention_layernorm.weight": "model-00078-of-00213.safetensors",
    "model.layers.31.self_attn.k_norm.weight": "model-00078-of-00213.safetensors",
    "model.layers.31.self_attn.k_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.31.self_attn.o_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.31.self_attn.q_norm.weight": "model-00078-of-00213.safetensors",
    "model.layers.31.self_attn.q_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.31.self_attn.v_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.32.input_layernorm.weight": "model-00078-of-00213.safetensors",
    "model.layers.32.linear_attn.A_log": "model-00078-of-00213.safetensors",
    "model.layers.32.linear_attn.conv1d.weight": "model-00078-of-00213.safetensors",
    "model.layers.32.linear_attn.dt_bias": "model-00078-of-00213.safetensors",
    "model.layers.32.linear_attn.in_proj_a.weight": "model-00078-of-00213.safetensors",
    "model.layers.32.linear_attn.in_proj_b.weight": "model-00078-of-00213.safetensors",
    "model.layers.32.linear_attn.in_proj_qkv.weight": "model-00078-of-00213.safetensors",
    "model.layers.32.linear_attn.in_proj_z.weight": "model-00078-of-00213.safetensors",
    "model.layers.32.linear_attn.norm.weight": "model-00078-of-00213.safetensors",
    "model.layers.32.linear_attn.out_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.32.mlp.experts.down_proj": "model-00075-of-00213.safetensors",
    "model.layers.32.mlp.experts.gate_up_proj": "model-00074-of-00213.safetensors",
    "model.layers.32.mlp.gate.weight": "model-00078-of-00213.safetensors",
    "model.layers.32.mlp.shared_expert.down_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.32.mlp.shared_expert.gate_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.32.mlp.shared_expert.up_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.32.mlp.shared_expert_gate.weight": "model-00078-of-00213.safetensors",
    "model.layers.32.post_attention_layernorm.weight": "model-00078-of-00213.safetensors",
    "model.layers.33.input_layernorm.weight": "model-00078-of-00213.safetensors",
    "model.layers.33.linear_attn.A_log": "model-00078-of-00213.safetensors",
    "model.layers.33.linear_attn.conv1d.weight": "model-00078-of-00213.safetensors",
    "model.layers.33.linear_attn.dt_bias": "model-00078-of-00213.safetensors",
    "model.layers.33.linear_attn.in_proj_a.weight": "model-00078-of-00213.safetensors",
    "model.layers.33.linear_attn.in_proj_b.weight": "model-00078-of-00213.safetensors",
    "model.layers.33.linear_attn.in_proj_qkv.weight": "model-00078-of-00213.safetensors",
    "model.layers.33.linear_attn.in_proj_z.weight": "model-00078-of-00213.safetensors",
    "model.layers.33.linear_attn.norm.weight": "model-00078-of-00213.safetensors",
    "model.layers.33.linear_attn.out_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.33.mlp.experts.down_proj": "model-00077-of-00213.safetensors",
    "model.layers.33.mlp.experts.gate_up_proj": "model-00076-of-00213.safetensors",
    "model.layers.33.mlp.gate.weight": "model-00078-of-00213.safetensors",
    "model.layers.33.mlp.shared_expert.down_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.33.mlp.shared_expert.gate_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.33.mlp.shared_expert.up_proj.weight": "model-00078-of-00213.safetensors",
    "model.layers.33.mlp.shared_expert_gate.weight": "model-00078-of-00213.safetensors",
    "model.layers.33.post_attention_layernorm.weight": "model-00078-of-00213.safetensors",
    "model.layers.34.input_layernorm.weight": "model-00087-of-00213.safetensors",
    "model.layers.34.linear_attn.A_log": "model-00087-of-00213.safetensors",
    "model.layers.34.linear_attn.conv1d.weight": "model-00087-of-00213.safetensors",
    "model.layers.34.linear_attn.dt_bias": "model-00087-of-00213.safetensors",
    "model.layers.34.linear_attn.in_proj_a.weight": "model-00087-of-00213.safetensors",
    "model.layers.34.linear_attn.in_proj_b.weight": "model-00087-of-00213.safetensors",
    "model.layers.34.linear_attn.in_proj_qkv.weight": "model-00087-of-00213.safetensors",
    "model.layers.34.linear_attn.in_proj_z.weight": "model-00087-of-00213.safetensors",
    "model.layers.34.linear_attn.norm.weight": "model-00087-of-00213.safetensors",
    "model.layers.34.linear_attn.out_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.34.mlp.experts.down_proj": "model-00080-of-00213.safetensors",
    "model.layers.34.mlp.experts.gate_up_proj": "model-00079-of-00213.safetensors",
    "model.layers.34.mlp.gate.weight": "model-00087-of-00213.safetensors",
    "model.layers.34.mlp.shared_expert.down_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.34.mlp.shared_expert.gate_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.34.mlp.shared_expert.up_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.34.mlp.shared_expert_gate.weight": "model-00087-of-00213.safetensors",
    "model.layers.34.post_attention_layernorm.weight": "model-00087-of-00213.safetensors",
    "model.layers.35.input_layernorm.weight": "model-00087-of-00213.safetensors",
    "model.layers.35.mlp.experts.down_proj": "model-00082-of-00213.safetensors",
    "model.layers.35.mlp.experts.gate_up_proj": "model-00081-of-00213.safetensors",
    "model.layers.35.mlp.gate.weight": "model-00087-of-00213.safetensors",
    "model.layers.35.mlp.shared_expert.down_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.35.mlp.shared_expert.gate_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.35.mlp.shared_expert.up_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.35.mlp.shared_expert_gate.weight": "model-00087-of-00213.safetensors",
    "model.layers.35.post_attention_layernorm.weight": "model-00087-of-00213.safetensors",
    "model.layers.35.self_attn.k_norm.weight": "model-00087-of-00213.safetensors",
    "model.layers.35.self_attn.k_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.35.self_attn.o_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.35.self_attn.q_norm.weight": "model-00087-of-00213.safetensors",
    "model.layers.35.self_attn.q_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.35.self_attn.v_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.36.input_layernorm.weight": "model-00087-of-00213.safetensors",
    "model.layers.36.linear_attn.A_log": "model-00087-of-00213.safetensors",
    "model.layers.36.linear_attn.conv1d.weight": "model-00087-of-00213.safetensors",
    "model.layers.36.linear_attn.dt_bias": "model-00087-of-00213.safetensors",
    "model.layers.36.linear_attn.in_proj_a.weight": "model-00087-of-00213.safetensors",
    "model.layers.36.linear_attn.in_proj_b.weight": "model-00087-of-00213.safetensors",
    "model.layers.36.linear_attn.in_proj_qkv.weight": "model-00087-of-00213.safetensors",
    "model.layers.36.linear_attn.in_proj_z.weight": "model-00087-of-00213.safetensors",
    "model.layers.36.linear_attn.norm.weight": "model-00087-of-00213.safetensors",
    "model.layers.36.linear_attn.out_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.36.mlp.experts.down_proj": "model-00084-of-00213.safetensors",
    "model.layers.36.mlp.experts.gate_up_proj": "model-00083-of-00213.safetensors",
    "model.layers.36.mlp.gate.weight": "model-00087-of-00213.safetensors",
    "model.layers.36.mlp.shared_expert.down_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.36.mlp.shared_expert.gate_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.36.mlp.shared_expert.up_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.36.mlp.shared_expert_gate.weight": "model-00087-of-00213.safetensors",
    "model.layers.36.post_attention_layernorm.weight": "model-00087-of-00213.safetensors",
    "model.layers.37.input_layernorm.weight": "model-00087-of-00213.safetensors",
    "model.layers.37.linear_attn.A_log": "model-00087-of-00213.safetensors",
    "model.layers.37.linear_attn.conv1d.weight": "model-00087-of-00213.safetensors",
    "model.layers.37.linear_attn.dt_bias": "model-00087-of-00213.safetensors",
    "model.layers.37.linear_attn.in_proj_a.weight": "model-00087-of-00213.safetensors",
    "model.layers.37.linear_attn.in_proj_b.weight": "model-00087-of-00213.safetensors",
    "model.layers.37.linear_attn.in_proj_qkv.weight": "model-00087-of-00213.safetensors",
    "model.layers.37.linear_attn.in_proj_z.weight": "model-00087-of-00213.safetensors",
    "model.layers.37.linear_attn.norm.weight": "model-00087-of-00213.safetensors",
    "model.layers.37.linear_attn.out_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.37.mlp.experts.down_proj": "model-00086-of-00213.safetensors",
    "model.layers.37.mlp.experts.gate_up_proj": "model-00085-of-00213.safetensors",
    "model.layers.37.mlp.gate.weight": "model-00087-of-00213.safetensors",
    "model.layers.37.mlp.shared_expert.down_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.37.mlp.shared_expert.gate_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.37.mlp.shared_expert.up_proj.weight": "model-00087-of-00213.safetensors",
    "model.layers.37.mlp.shared_expert_gate.weight": "model-00087-of-00213.safetensors",
    "model.layers.37.post_attention_layernorm.weight": "model-00087-of-00213.safetensors",
    "model.layers.38.input_layernorm.weight": "model-00087-of-00213.safetensors",
    "model.layers.38.linear_attn.A_log": "model-00087-of-00213.safetensors",
    "model.layers.38.linear_attn.conv1d.weight": "model-00087-of-00213.safetensors",
    "model.layers.38.linear_attn.dt_bias": "model-00087-of-00213.safetensors",
    "model.layers.38.linear_attn.in_proj_a.weight": "model-00087-of-00213.safetensors",
    "model.layers.38.linear_attn.in_proj_b.weight": "model-00087-of-00213.safetensors",
    "model.layers.38.linear_attn.in_proj_qkv.weight": "model-00096-of-00213.safetensors",
    "model.layers.38.linear_attn.in_proj_z.weight": "model-00096-of-00213.safetensors",
    "model.layers.38.linear_attn.norm.weight": "model-00096-of-00213.safetensors",
    "model.layers.38.linear_attn.out_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.38.mlp.experts.down_proj": "model-00089-of-00213.safetensors",
    "model.layers.38.mlp.experts.gate_up_proj": "model-00088-of-00213.safetensors",
    "model.layers.38.mlp.gate.weight": "model-00096-of-00213.safetensors",
    "model.layers.38.mlp.shared_expert.down_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.38.mlp.shared_expert.gate_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.38.mlp.shared_expert.up_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.38.mlp.shared_expert_gate.weight": "model-00096-of-00213.safetensors",
    "model.layers.38.post_attention_layernorm.weight": "model-00096-of-00213.safetensors",
    "model.layers.39.input_layernorm.weight": "model-00096-of-00213.safetensors",
    "model.layers.39.mlp.experts.down_proj": "model-00091-of-00213.safetensors",
    "model.layers.39.mlp.experts.gate_up_proj": "model-00090-of-00213.safetensors",
    "model.layers.39.mlp.gate.weight": "model-00096-of-00213.safetensors",
    "model.layers.39.mlp.shared_expert.down_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.39.mlp.shared_expert.gate_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.39.mlp.shared_expert.up_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.39.mlp.shared_expert_gate.weight": "model-00096-of-00213.safetensors",
    "model.layers.39.post_attention_layernorm.weight": "model-00096-of-00213.safetensors",
    "model.layers.39.self_attn.k_norm.weight": "model-00096-of-00213.safetensors",
    "model.layers.39.self_attn.k_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.39.self_attn.o_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.39.self_attn.q_norm.weight": "model-00096-of-00213.safetensors",
    "model.layers.39.self_attn.q_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.39.self_attn.v_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.4.input_layernorm.weight": "model-00020-of-00213.safetensors",
    "model.layers.4.linear_attn.A_log": "model-00020-of-00213.safetensors",
    "model.layers.4.linear_attn.conv1d.weight": "model-00020-of-00213.safetensors",
    "model.layers.4.linear_attn.dt_bias": "model-00020-of-00213.safetensors",
    "model.layers.4.linear_attn.in_proj_a.weight": "model-00020-of-00213.safetensors",
    "model.layers.4.linear_attn.in_proj_b.weight": "model-00020-of-00213.safetensors",
    "model.layers.4.linear_attn.in_proj_qkv.weight": "model-00020-of-00213.safetensors",
    "model.layers.4.linear_attn.in_proj_z.weight": "model-00020-of-00213.safetensors",
    "model.layers.4.linear_attn.norm.weight": "model-00020-of-00213.safetensors",
    "model.layers.4.linear_attn.out_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.4.mlp.experts.down_proj": "model-00013-of-00213.safetensors",
    "model.layers.4.mlp.experts.gate_up_proj": "model-00012-of-00213.safetensors",
    "model.layers.4.mlp.gate.weight": "model-00020-of-00213.safetensors",
    "model.layers.4.mlp.shared_expert.down_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.4.mlp.shared_expert.gate_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.4.mlp.shared_expert.up_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.4.mlp.shared_expert_gate.weight": "model-00020-of-00213.safetensors",
    "model.layers.4.post_attention_layernorm.weight": "model-00020-of-00213.safetensors",
    "model.layers.40.input_layernorm.weight": "model-00096-of-00213.safetensors",
    "model.layers.40.linear_attn.A_log": "model-00096-of-00213.safetensors",
    "model.layers.40.linear_attn.conv1d.weight": "model-00096-of-00213.safetensors",
    "model.layers.40.linear_attn.dt_bias": "model-00096-of-00213.safetensors",
    "model.layers.40.linear_attn.in_proj_a.weight": "model-00096-of-00213.safetensors",
    "model.layers.40.linear_attn.in_proj_b.weight": "model-00096-of-00213.safetensors",
    "model.layers.40.linear_attn.in_proj_qkv.weight": "model-00096-of-00213.safetensors",
    "model.layers.40.linear_attn.in_proj_z.weight": "model-00096-of-00213.safetensors",
    "model.layers.40.linear_attn.norm.weight": "model-00096-of-00213.safetensors",
    "model.layers.40.linear_attn.out_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.40.mlp.experts.down_proj": "model-00093-of-00213.safetensors",
    "model.layers.40.mlp.experts.gate_up_proj": "model-00092-of-00213.safetensors",
    "model.layers.40.mlp.gate.weight": "model-00096-of-00213.safetensors",
    "model.layers.40.mlp.shared_expert.down_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.40.mlp.shared_expert.gate_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.40.mlp.shared_expert.up_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.40.mlp.shared_expert_gate.weight": "model-00096-of-00213.safetensors",
    "model.layers.40.post_attention_layernorm.weight": "model-00096-of-00213.safetensors",
    "model.layers.41.input_layernorm.weight": "model-00096-of-00213.safetensors",
    "model.layers.41.linear_attn.A_log": "model-00096-of-00213.safetensors",
    "model.layers.41.linear_attn.conv1d.weight": "model-00096-of-00213.safetensors",
    "model.layers.41.linear_attn.dt_bias": "model-00096-of-00213.safetensors",
    "model.layers.41.linear_attn.in_proj_a.weight": "model-00096-of-00213.safetensors",
    "model.layers.41.linear_attn.in_proj_b.weight": "model-00096-of-00213.safetensors",
    "model.layers.41.linear_attn.in_proj_qkv.weight": "model-00096-of-00213.safetensors",
    "model.layers.41.linear_attn.in_proj_z.weight": "model-00096-of-00213.safetensors",
    "model.layers.41.linear_attn.norm.weight": "model-00096-of-00213.safetensors",
    "model.layers.41.linear_attn.out_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.41.mlp.experts.down_proj": "model-00095-of-00213.safetensors",
    "model.layers.41.mlp.experts.gate_up_proj": "model-00094-of-00213.safetensors",
    "model.layers.41.mlp.gate.weight": "model-00096-of-00213.safetensors",
    "model.layers.41.mlp.shared_expert.down_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.41.mlp.shared_expert.gate_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.41.mlp.shared_expert.up_proj.weight": "model-00096-of-00213.safetensors",
    "model.layers.41.mlp.shared_expert_gate.weight": "model-00096-of-00213.safetensors",
    "model.layers.41.post_attention_layernorm.weight": "model-00096-of-00213.safetensors",
    "model.layers.42.input_layernorm.weight": "model-00096-of-00213.safetensors",
    "model.layers.42.linear_attn.A_log": "model-00096-of-00213.safetensors",
    "model.layers.42.linear_attn.conv1d.weight": "model-00096-of-00213.safetensors",
    "model.layers.42.linear_attn.dt_bias": "model-00096-of-00213.safetensors",
    "model.layers.42.linear_attn.in_proj_a.weight": "model-00096-of-00213.safetensors",
    "model.layers.42.linear_attn.in_proj_b.weight": "model-00096-of-00213.safetensors",
    "model.layers.42.linear_attn.in_proj_qkv.weight": "model-00105-of-00213.safetensors",
    "model.layers.42.linear_attn.in_proj_z.weight": "model-00105-of-00213.safetensors",
    "model.layers.42.linear_attn.norm.weight": "model-00105-of-00213.safetensors",
    "model.layers.42.linear_attn.out_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.42.mlp.experts.down_proj": "model-00098-of-00213.safetensors",
    "model.layers.42.mlp.experts.gate_up_proj": "model-00097-of-00213.safetensors",
    "model.layers.42.mlp.gate.weight": "model-00105-of-00213.safetensors",
    "model.layers.42.mlp.shared_expert.down_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.42.mlp.shared_expert.gate_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.42.mlp.shared_expert.up_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.42.mlp.shared_expert_gate.weight": "model-00105-of-00213.safetensors",
    "model.layers.42.post_attention_layernorm.weight": "model-00105-of-00213.safetensors",
    "model.layers.43.input_layernorm.weight": "model-00105-of-00213.safetensors",
    "model.layers.43.mlp.experts.down_proj": "model-00100-of-00213.safetensors",
    "model.layers.43.mlp.experts.gate_up_proj": "model-00099-of-00213.safetensors",
    "model.layers.43.mlp.gate.weight": "model-00105-of-00213.safetensors",
    "model.layers.43.mlp.shared_expert.down_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.43.mlp.shared_expert.gate_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.43.mlp.shared_expert.up_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.43.mlp.shared_expert_gate.weight": "model-00105-of-00213.safetensors",
    "model.layers.43.post_attention_layernorm.weight": "model-00105-of-00213.safetensors",
    "model.layers.43.self_attn.k_norm.weight": "model-00105-of-00213.safetensors",
    "model.layers.43.self_attn.k_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.43.self_attn.o_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.43.self_attn.q_norm.weight": "model-00105-of-00213.safetensors",
    "model.layers.43.self_attn.q_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.43.self_attn.v_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.44.input_layernorm.weight": "model-00105-of-00213.safetensors",
    "model.layers.44.linear_attn.A_log": "model-00105-of-00213.safetensors",
    "model.layers.44.linear_attn.conv1d.weight": "model-00105-of-00213.safetensors",
    "model.layers.44.linear_attn.dt_bias": "model-00105-of-00213.safetensors",
    "model.layers.44.linear_attn.in_proj_a.weight": "model-00105-of-00213.safetensors",
    "model.layers.44.linear_attn.in_proj_b.weight": "model-00105-of-00213.safetensors",
    "model.layers.44.linear_attn.in_proj_qkv.weight": "model-00105-of-00213.safetensors",
    "model.layers.44.linear_attn.in_proj_z.weight": "model-00105-of-00213.safetensors",
    "model.layers.44.linear_attn.norm.weight": "model-00105-of-00213.safetensors",
    "model.layers.44.linear_attn.out_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.44.mlp.experts.down_proj": "model-00102-of-00213.safetensors",
    "model.layers.44.mlp.experts.gate_up_proj": "model-00101-of-00213.safetensors",
    "model.layers.44.mlp.gate.weight": "model-00105-of-00213.safetensors",
    "model.layers.44.mlp.shared_expert.down_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.44.mlp.shared_expert.gate_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.44.mlp.shared_expert.up_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.44.mlp.shared_expert_gate.weight": "model-00105-of-00213.safetensors",
    "model.layers.44.post_attention_layernorm.weight": "model-00105-of-00213.safetensors",
    "model.layers.45.input_layernorm.weight": "model-00105-of-00213.safetensors",
    "model.layers.45.linear_attn.A_log": "model-00105-of-00213.safetensors",
    "model.layers.45.linear_attn.conv1d.weight": "model-00105-of-00213.safetensors",
    "model.layers.45.linear_attn.dt_bias": "model-00105-of-00213.safetensors",
    "model.layers.45.linear_attn.in_proj_a.weight": "model-00105-of-00213.safetensors",
    "model.layers.45.linear_attn.in_proj_b.weight": "model-00105-of-00213.safetensors",
    "model.layers.45.linear_attn.in_proj_qkv.weight": "model-00105-of-00213.safetensors",
    "model.layers.45.linear_attn.in_proj_z.weight": "model-00105-of-00213.safetensors",
    "model.layers.45.linear_attn.norm.weight": "model-00105-of-00213.safetensors",
    "model.layers.45.linear_attn.out_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.45.mlp.experts.down_proj": "model-00104-of-00213.safetensors",
    "model.layers.45.mlp.experts.gate_up_proj": "model-00103-of-00213.safetensors",
    "model.layers.45.mlp.gate.weight": "model-00105-of-00213.safetensors",
    "model.layers.45.mlp.shared_expert.down_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.45.mlp.shared_expert.gate_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.45.mlp.shared_expert.up_proj.weight": "model-00105-of-00213.safetensors",
    "model.layers.45.mlp.shared_expert_gate.weight": "model-00105-of-00213.safetensors",
    "model.layers.45.post_attention_layernorm.weight": "model-00105-of-00213.safetensors",
    "model.layers.46.input_layernorm.weight": "model-00114-of-00213.safetensors",
    "model.layers.46.linear_attn.A_log": "model-00114-of-00213.safetensors",
    "model.layers.46.linear_attn.conv1d.weight": "model-00114-of-00213.safetensors",
    "model.layers.46.linear_attn.dt_bias": "model-00114-of-00213.safetensors",
    "model.layers.46.linear_attn.in_proj_a.weight": "model-00114-of-00213.safetensors",
    "model.layers.46.linear_attn.in_proj_b.weight": "model-00114-of-00213.safetensors",
    "model.layers.46.linear_attn.in_proj_qkv.weight": "model-00114-of-00213.safetensors",
    "model.layers.46.linear_attn.in_proj_z.weight": "model-00114-of-00213.safetensors",
    "model.layers.46.linear_attn.norm.weight": "model-00114-of-00213.safetensors",
    "model.layers.46.linear_attn.out_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.46.mlp.experts.down_proj": "model-00107-of-00213.safetensors",
    "model.layers.46.mlp.experts.gate_up_proj": "model-00106-of-00213.safetensors",
    "model.layers.46.mlp.gate.weight": "model-00114-of-00213.safetensors",
    "model.layers.46.mlp.shared_expert.down_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.46.mlp.shared_expert.gate_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.46.mlp.shared_expert.up_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.46.mlp.shared_expert_gate.weight": "model-00114-of-00213.safetensors",
    "model.layers.46.post_attention_layernorm.weight": "model-00114-of-00213.safetensors",
    "model.layers.47.input_layernorm.weight": "model-00114-of-00213.safetensors",
    "model.layers.47.mlp.experts.down_proj": "model-00109-of-00213.safetensors",
    "model.layers.47.mlp.experts.gate_up_proj": "model-00108-of-00213.safetensors",
    "model.layers.47.mlp.gate.weight": "model-00114-of-00213.safetensors",
    "model.layers.47.mlp.shared_expert.down_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.47.mlp.shared_expert.gate_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.47.mlp.shared_expert.up_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.47.mlp.shared_expert_gate.weight": "model-00114-of-00213.safetensors",
    "model.layers.47.post_attention_layernorm.weight": "model-00114-of-00213.safetensors",
    "model.layers.47.self_attn.k_norm.weight": "model-00114-of-00213.safetensors",
    "model.layers.47.self_attn.k_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.47.self_attn.o_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.47.self_attn.q_norm.weight": "model-00114-of-00213.safetensors",
    "model.layers.47.self_attn.q_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.47.self_attn.v_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.48.input_layernorm.weight": "model-00114-of-00213.safetensors",
    "model.layers.48.linear_attn.A_log": "model-00114-of-00213.safetensors",
    "model.layers.48.linear_attn.conv1d.weight": "model-00114-of-00213.safetensors",
    "model.layers.48.linear_attn.dt_bias": "model-00114-of-00213.safetensors",
    "model.layers.48.linear_attn.in_proj_a.weight": "model-00114-of-00213.safetensors",
    "model.layers.48.linear_attn.in_proj_b.weight": "model-00114-of-00213.safetensors",
    "model.layers.48.linear_attn.in_proj_qkv.weight": "model-00114-of-00213.safetensors",
    "model.layers.48.linear_attn.in_proj_z.weight": "model-00114-of-00213.safetensors",
    "model.layers.48.linear_attn.norm.weight": "model-00114-of-00213.safetensors",
    "model.layers.48.linear_attn.out_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.48.mlp.experts.down_proj": "model-00111-of-00213.safetensors",
    "model.layers.48.mlp.experts.gate_up_proj": "model-00110-of-00213.safetensors",
    "model.layers.48.mlp.gate.weight": "model-00114-of-00213.safetensors",
    "model.layers.48.mlp.shared_expert.down_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.48.mlp.shared_expert.gate_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.48.mlp.shared_expert.up_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.48.mlp.shared_expert_gate.weight": "model-00114-of-00213.safetensors",
    "model.layers.48.post_attention_layernorm.weight": "model-00114-of-00213.safetensors",
    "model.layers.49.input_layernorm.weight": "model-00114-of-00213.safetensors",
    "model.layers.49.linear_attn.A_log": "model-00114-of-00213.safetensors",
    "model.layers.49.linear_attn.conv1d.weight": "model-00114-of-00213.safetensors",
    "model.layers.49.linear_attn.dt_bias": "model-00114-of-00213.safetensors",
    "model.layers.49.linear_attn.in_proj_a.weight": "model-00114-of-00213.safetensors",
    "model.layers.49.linear_attn.in_proj_b.weight": "model-00114-of-00213.safetensors",
    "model.layers.49.linear_attn.in_proj_qkv.weight": "model-00114-of-00213.safetensors",
    "model.layers.49.linear_attn.in_proj_z.weight": "model-00114-of-00213.safetensors",
    "model.layers.49.linear_attn.norm.weight": "model-00114-of-00213.safetensors",
    "model.layers.49.linear_attn.out_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.49.mlp.experts.down_proj": "model-00113-of-00213.safetensors",
    "model.layers.49.mlp.experts.gate_up_proj": "model-00112-of-00213.safetensors",
    "model.layers.49.mlp.gate.weight": "model-00114-of-00213.safetensors",
    "model.layers.49.mlp.shared_expert.down_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.49.mlp.shared_expert.gate_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.49.mlp.shared_expert.up_proj.weight": "model-00114-of-00213.safetensors",
    "model.layers.49.mlp.shared_expert_gate.weight": "model-00114-of-00213.safetensors",
    "model.layers.49.post_attention_layernorm.weight": "model-00114-of-00213.safetensors",
    "model.layers.5.input_layernorm.weight": "model-00020-of-00213.safetensors",
    "model.layers.5.linear_attn.A_log": "model-00020-of-00213.safetensors",
    "model.layers.5.linear_attn.conv1d.weight": "model-00020-of-00213.safetensors",
    "model.layers.5.linear_attn.dt_bias": "model-00020-of-00213.safetensors",
    "model.layers.5.linear_attn.in_proj_a.weight": "model-00020-of-00213.safetensors",
    "model.layers.5.linear_attn.in_proj_b.weight": "model-00020-of-00213.safetensors",
    "model.layers.5.linear_attn.in_proj_qkv.weight": "model-00020-of-00213.safetensors",
    "model.layers.5.linear_attn.in_proj_z.weight": "model-00020-of-00213.safetensors",
    "model.layers.5.linear_attn.norm.weight": "model-00020-of-00213.safetensors",
    "model.layers.5.linear_attn.out_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.5.mlp.experts.down_proj": "model-00015-of-00213.safetensors",
    "model.layers.5.mlp.experts.gate_up_proj": "model-00014-of-00213.safetensors",
    "model.layers.5.mlp.gate.weight": "model-00020-of-00213.safetensors",
    "model.layers.5.mlp.shared_expert.down_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.5.mlp.shared_expert.gate_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.5.mlp.shared_expert.up_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.5.mlp.shared_expert_gate.weight": "model-00020-of-00213.safetensors",
    "model.layers.5.post_attention_layernorm.weight": "model-00020-of-00213.safetensors",
    "model.layers.50.input_layernorm.weight": "model-00114-of-00213.safetensors",
    "model.layers.50.linear_attn.A_log": "model-00114-of-00213.safetensors",
    "model.layers.50.linear_attn.conv1d.weight": "model-00114-of-00213.safetensors",
    "model.layers.50.linear_attn.dt_bias": "model-00114-of-00213.safetensors",
    "model.layers.50.linear_attn.in_proj_a.weight": "model-00114-of-00213.safetensors",
    "model.layers.50.linear_attn.in_proj_b.weight": "model-00114-of-00213.safetensors",
    "model.layers.50.linear_attn.in_proj_qkv.weight": "model-00123-of-00213.safetensors",
    "model.layers.50.linear_attn.in_proj_z.weight": "model-00123-of-00213.safetensors",
    "model.layers.50.linear_attn.norm.weight": "model-00123-of-00213.safetensors",
    "model.layers.50.linear_attn.out_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.50.mlp.experts.down_proj": "model-00116-of-00213.safetensors",
    "model.layers.50.mlp.experts.gate_up_proj": "model-00115-of-00213.safetensors",
    "model.layers.50.mlp.gate.weight": "model-00123-of-00213.safetensors",
    "model.layers.50.mlp.shared_expert.down_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.50.mlp.shared_expert.gate_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.50.mlp.shared_expert.up_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.50.mlp.shared_expert_gate.weight": "model-00123-of-00213.safetensors",
    "model.layers.50.post_attention_layernorm.weight": "model-00123-of-00213.safetensors",
    "model.layers.51.input_layernorm.weight": "model-00123-of-00213.safetensors",
    "model.layers.51.mlp.experts.down_proj": "model-00118-of-00213.safetensors",
    "model.layers.51.mlp.experts.gate_up_proj": "model-00117-of-00213.safetensors",
    "model.layers.51.mlp.gate.weight": "model-00123-of-00213.safetensors",
    "model.layers.51.mlp.shared_expert.down_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.51.mlp.shared_expert.gate_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.51.mlp.shared_expert.up_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.51.mlp.shared_expert_gate.weight": "model-00123-of-00213.safetensors",
    "model.layers.51.post_attention_layernorm.weight": "model-00123-of-00213.safetensors",
    "model.layers.51.self_attn.k_norm.weight": "model-00123-of-00213.safetensors",
    "model.layers.51.self_attn.k_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.51.self_attn.o_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.51.self_attn.q_norm.weight": "model-00123-of-00213.safetensors",
    "model.layers.51.self_attn.q_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.51.self_attn.v_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.52.input_layernorm.weight": "model-00123-of-00213.safetensors",
    "model.layers.52.linear_attn.A_log": "model-00123-of-00213.safetensors",
    "model.layers.52.linear_attn.conv1d.weight": "model-00123-of-00213.safetensors",
    "model.layers.52.linear_attn.dt_bias": "model-00123-of-00213.safetensors",
    "model.layers.52.linear_attn.in_proj_a.weight": "model-00123-of-00213.safetensors",
    "model.layers.52.linear_attn.in_proj_b.weight": "model-00123-of-00213.safetensors",
    "model.layers.52.linear_attn.in_proj_qkv.weight": "model-00123-of-00213.safetensors",
    "model.layers.52.linear_attn.in_proj_z.weight": "model-00123-of-00213.safetensors",
    "model.layers.52.linear_attn.norm.weight": "model-00123-of-00213.safetensors",
    "model.layers.52.linear_attn.out_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.52.mlp.experts.down_proj": "model-00120-of-00213.safetensors",
    "model.layers.52.mlp.experts.gate_up_proj": "model-00119-of-00213.safetensors",
    "model.layers.52.mlp.gate.weight": "model-00123-of-00213.safetensors",
    "model.layers.52.mlp.shared_expert.down_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.52.mlp.shared_expert.gate_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.52.mlp.shared_expert.up_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.52.mlp.shared_expert_gate.weight": "model-00123-of-00213.safetensors",
    "model.layers.52.post_attention_layernorm.weight": "model-00123-of-00213.safetensors",
    "model.layers.53.input_layernorm.weight": "model-00123-of-00213.safetensors",
    "model.layers.53.linear_attn.A_log": "model-00123-of-00213.safetensors",
    "model.layers.53.linear_attn.conv1d.weight": "model-00123-of-00213.safetensors",
    "model.layers.53.linear_attn.dt_bias": "model-00123-of-00213.safetensors",
    "model.layers.53.linear_attn.in_proj_a.weight": "model-00123-of-00213.safetensors",
    "model.layers.53.linear_attn.in_proj_b.weight": "model-00123-of-00213.safetensors",
    "model.layers.53.linear_attn.in_proj_qkv.weight": "model-00123-of-00213.safetensors",
    "model.layers.53.linear_attn.in_proj_z.weight": "model-00123-of-00213.safetensors",
    "model.layers.53.linear_attn.norm.weight": "model-00123-of-00213.safetensors",
    "model.layers.53.linear_attn.out_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.53.mlp.experts.down_proj": "model-00122-of-00213.safetensors",
    "model.layers.53.mlp.experts.gate_up_proj": "model-00121-of-00213.safetensors",
    "model.layers.53.mlp.gate.weight": "model-00123-of-00213.safetensors",
    "model.layers.53.mlp.shared_expert.down_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.53.mlp.shared_expert.gate_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.53.mlp.shared_expert.up_proj.weight": "model-00123-of-00213.safetensors",
    "model.layers.53.mlp.shared_expert_gate.weight": "model-00123-of-00213.safetensors",
    "model.layers.53.post_attention_layernorm.weight": "model-00123-of-00213.safetensors",
    "model.layers.54.input_layernorm.weight": "model-00123-of-00213.safetensors",
    "model.layers.54.linear_attn.A_log": "model-00123-of-00213.safetensors",
    "model.layers.54.linear_attn.conv1d.weight": "model-00123-of-00213.safetensors",
    "model.layers.54.linear_attn.dt_bias": "model-00123-of-00213.safetensors",
    "model.layers.54.linear_attn.in_proj_a.weight": "model-00123-of-00213.safetensors",
    "model.layers.54.linear_attn.in_proj_b.weight": "model-00123-of-00213.safetensors",
    "model.layers.54.linear_attn.in_proj_qkv.weight": "model-00132-of-00213.safetensors",
    "model.layers.54.linear_attn.in_proj_z.weight": "model-00132-of-00213.safetensors",
    "model.layers.54.linear_attn.norm.weight": "model-00132-of-00213.safetensors",
    "model.layers.54.linear_attn.out_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.54.mlp.experts.down_proj": "model-00125-of-00213.safetensors",
    "model.layers.54.mlp.experts.gate_up_proj": "model-00124-of-00213.safetensors",
    "model.layers.54.mlp.gate.weight": "model-00132-of-00213.safetensors",
    "model.layers.54.mlp.shared_expert.down_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.54.mlp.shared_expert.gate_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.54.mlp.shared_expert.up_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.54.mlp.shared_expert_gate.weight": "model-00132-of-00213.safetensors",
    "model.layers.54.post_attention_layernorm.weight": "model-00132-of-00213.safetensors",
    "model.layers.55.input_layernorm.weight": "model-00132-of-00213.safetensors",
    "model.layers.55.mlp.experts.down_proj": "model-00127-of-00213.safetensors",
    "model.layers.55.mlp.experts.gate_up_proj": "model-00126-of-00213.safetensors",
    "model.layers.55.mlp.gate.weight": "model-00132-of-00213.safetensors",
    "model.layers.55.mlp.shared_expert.down_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.55.mlp.shared_expert.gate_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.55.mlp.shared_expert.up_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.55.mlp.shared_expert_gate.weight": "model-00132-of-00213.safetensors",
    "model.layers.55.post_attention_layernorm.weight": "model-00132-of-00213.safetensors",
    "model.layers.55.self_attn.k_norm.weight": "model-00132-of-00213.safetensors",
    "model.layers.55.self_attn.k_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.55.self_attn.o_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.55.self_attn.q_norm.weight": "model-00132-of-00213.safetensors",
    "model.layers.55.self_attn.q_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.55.self_attn.v_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.56.input_layernorm.weight": "model-00132-of-00213.safetensors",
    "model.layers.56.linear_attn.A_log": "model-00132-of-00213.safetensors",
    "model.layers.56.linear_attn.conv1d.weight": "model-00132-of-00213.safetensors",
    "model.layers.56.linear_attn.dt_bias": "model-00132-of-00213.safetensors",
    "model.layers.56.linear_attn.in_proj_a.weight": "model-00132-of-00213.safetensors",
    "model.layers.56.linear_attn.in_proj_b.weight": "model-00132-of-00213.safetensors",
    "model.layers.56.linear_attn.in_proj_qkv.weight": "model-00132-of-00213.safetensors",
    "model.layers.56.linear_attn.in_proj_z.weight": "model-00132-of-00213.safetensors",
    "model.layers.56.linear_attn.norm.weight": "model-00132-of-00213.safetensors",
    "model.layers.56.linear_attn.out_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.56.mlp.experts.down_proj": "model-00129-of-00213.safetensors",
    "model.layers.56.mlp.experts.gate_up_proj": "model-00128-of-00213.safetensors",
    "model.layers.56.mlp.gate.weight": "model-00132-of-00213.safetensors",
    "model.layers.56.mlp.shared_expert.down_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.56.mlp.shared_expert.gate_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.56.mlp.shared_expert.up_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.56.mlp.shared_expert_gate.weight": "model-00132-of-00213.safetensors",
    "model.layers.56.post_attention_layernorm.weight": "model-00132-of-00213.safetensors",
    "model.layers.57.input_layernorm.weight": "model-00132-of-00213.safetensors",
    "model.layers.57.linear_attn.A_log": "model-00132-of-00213.safetensors",
    "model.layers.57.linear_attn.conv1d.weight": "model-00132-of-00213.safetensors",
    "model.layers.57.linear_attn.dt_bias": "model-00132-of-00213.safetensors",
    "model.layers.57.linear_attn.in_proj_a.weight": "model-00132-of-00213.safetensors",
    "model.layers.57.linear_attn.in_proj_b.weight": "model-00132-of-00213.safetensors",
    "model.layers.57.linear_attn.in_proj_qkv.weight": "model-00132-of-00213.safetensors",
    "model.layers.57.linear_attn.in_proj_z.weight": "model-00132-of-00213.safetensors",
    "model.layers.57.linear_attn.norm.weight": "model-00132-of-00213.safetensors",
    "model.layers.57.linear_attn.out_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.57.mlp.experts.down_proj": "model-00131-of-00213.safetensors",
    "model.layers.57.mlp.experts.gate_up_proj": "model-00130-of-00213.safetensors",
    "model.layers.57.mlp.gate.weight": "model-00132-of-00213.safetensors",
    "model.layers.57.mlp.shared_expert.down_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.57.mlp.shared_expert.gate_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.57.mlp.shared_expert.up_proj.weight": "model-00132-of-00213.safetensors",
    "model.layers.57.mlp.shared_expert_gate.weight": "model-00132-of-00213.safetensors",
    "model.layers.57.post_attention_layernorm.weight": "model-00132-of-00213.safetensors",
    "model.layers.58.input_layernorm.weight": "model-00141-of-00213.safetensors",
    "model.layers.58.linear_attn.A_log": "model-00141-of-00213.safetensors",
    "model.layers.58.linear_attn.conv1d.weight": "model-00141-of-00213.safetensors",
    "model.layers.58.linear_attn.dt_bias": "model-00141-of-00213.safetensors",
    "model.layers.58.linear_attn.in_proj_a.weight": "model-00141-of-00213.safetensors",
    "model.layers.58.linear_attn.in_proj_b.weight": "model-00141-of-00213.safetensors",
    "model.layers.58.linear_attn.in_proj_qkv.weight": "model-00141-of-00213.safetensors",
    "model.layers.58.linear_attn.in_proj_z.weight": "model-00141-of-00213.safetensors",
    "model.layers.58.linear_attn.norm.weight": "model-00141-of-00213.safetensors",
    "model.layers.58.linear_attn.out_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.58.mlp.experts.down_proj": "model-00134-of-00213.safetensors",
    "model.layers.58.mlp.experts.gate_up_proj": "model-00133-of-00213.safetensors",
    "model.layers.58.mlp.gate.weight": "model-00141-of-00213.safetensors",
    "model.layers.58.mlp.shared_expert.down_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.58.mlp.shared_expert.gate_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.58.mlp.shared_expert.up_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.58.mlp.shared_expert_gate.weight": "model-00141-of-00213.safetensors",
    "model.layers.58.post_attention_layernorm.weight": "model-00141-of-00213.safetensors",
    "model.layers.59.input_layernorm.weight": "model-00141-of-00213.safetensors",
    "model.layers.59.mlp.experts.down_proj": "model-00136-of-00213.safetensors",
    "model.layers.59.mlp.experts.gate_up_proj": "model-00135-of-00213.safetensors",
    "model.layers.59.mlp.gate.weight": "model-00141-of-00213.safetensors",
    "model.layers.59.mlp.shared_expert.down_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.59.mlp.shared_expert.gate_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.59.mlp.shared_expert.up_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.59.mlp.shared_expert_gate.weight": "model-00141-of-00213.safetensors",
    "model.layers.59.post_attention_layernorm.weight": "model-00141-of-00213.safetensors",
    "model.layers.59.self_attn.k_norm.weight": "model-00141-of-00213.safetensors",
    "model.layers.59.self_attn.k_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.59.self_attn.o_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.59.self_attn.q_norm.weight": "model-00141-of-00213.safetensors",
    "model.layers.59.self_attn.q_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.59.self_attn.v_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.6.input_layernorm.weight": "model-00020-of-00213.safetensors",
    "model.layers.6.linear_attn.A_log": "model-00020-of-00213.safetensors",
    "model.layers.6.linear_attn.conv1d.weight": "model-00020-of-00213.safetensors",
    "model.layers.6.linear_attn.dt_bias": "model-00020-of-00213.safetensors",
    "model.layers.6.linear_attn.in_proj_a.weight": "model-00020-of-00213.safetensors",
    "model.layers.6.linear_attn.in_proj_b.weight": "model-00020-of-00213.safetensors",
    "model.layers.6.linear_attn.in_proj_qkv.weight": "model-00020-of-00213.safetensors",
    "model.layers.6.linear_attn.in_proj_z.weight": "model-00020-of-00213.safetensors",
    "model.layers.6.linear_attn.norm.weight": "model-00020-of-00213.safetensors",
    "model.layers.6.linear_attn.out_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.6.mlp.experts.down_proj": "model-00017-of-00213.safetensors",
    "model.layers.6.mlp.experts.gate_up_proj": "model-00016-of-00213.safetensors",
    "model.layers.6.mlp.gate.weight": "model-00020-of-00213.safetensors",
    "model.layers.6.mlp.shared_expert.down_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.6.mlp.shared_expert.gate_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.6.mlp.shared_expert.up_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.6.mlp.shared_expert_gate.weight": "model-00020-of-00213.safetensors",
    "model.layers.6.post_attention_layernorm.weight": "model-00020-of-00213.safetensors",
    "model.layers.60.input_layernorm.weight": "model-00141-of-00213.safetensors",
    "model.layers.60.linear_attn.A_log": "model-00141-of-00213.safetensors",
    "model.layers.60.linear_attn.conv1d.weight": "model-00141-of-00213.safetensors",
    "model.layers.60.linear_attn.dt_bias": "model-00141-of-00213.safetensors",
    "model.layers.60.linear_attn.in_proj_a.weight": "model-00141-of-00213.safetensors",
    "model.layers.60.linear_attn.in_proj_b.weight": "model-00141-of-00213.safetensors",
    "model.layers.60.linear_attn.in_proj_qkv.weight": "model-00141-of-00213.safetensors",
    "model.layers.60.linear_attn.in_proj_z.weight": "model-00141-of-00213.safetensors",
    "model.layers.60.linear_attn.norm.weight": "model-00141-of-00213.safetensors",
    "model.layers.60.linear_attn.out_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.60.mlp.experts.down_proj": "model-00138-of-00213.safetensors",
    "model.layers.60.mlp.experts.gate_up_proj": "model-00137-of-00213.safetensors",
    "model.layers.60.mlp.gate.weight": "model-00141-of-00213.safetensors",
    "model.layers.60.mlp.shared_expert.down_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.60.mlp.shared_expert.gate_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.60.mlp.shared_expert.up_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.60.mlp.shared_expert_gate.weight": "model-00141-of-00213.safetensors",
    "model.layers.60.post_attention_layernorm.weight": "model-00141-of-00213.safetensors",
    "model.layers.61.input_layernorm.weight": "model-00141-of-00213.safetensors",
    "model.layers.61.linear_attn.A_log": "model-00141-of-00213.safetensors",
    "model.layers.61.linear_attn.conv1d.weight": "model-00141-of-00213.safetensors",
    "model.layers.61.linear_attn.dt_bias": "model-00141-of-00213.safetensors",
    "model.layers.61.linear_attn.in_proj_a.weight": "model-00141-of-00213.safetensors",
    "model.layers.61.linear_attn.in_proj_b.weight": "model-00141-of-00213.safetensors",
    "model.layers.61.linear_attn.in_proj_qkv.weight": "model-00141-of-00213.safetensors",
    "model.layers.61.linear_attn.in_proj_z.weight": "model-00141-of-00213.safetensors",
    "model.layers.61.linear_attn.norm.weight": "model-00141-of-00213.safetensors",
    "model.layers.61.linear_attn.out_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.61.mlp.experts.down_proj": "model-00140-of-00213.safetensors",
    "model.layers.61.mlp.experts.gate_up_proj": "model-00139-of-00213.safetensors",
    "model.layers.61.mlp.gate.weight": "model-00141-of-00213.safetensors",
    "model.layers.61.mlp.shared_expert.down_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.61.mlp.shared_expert.gate_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.61.mlp.shared_expert.up_proj.weight": "model-00141-of-00213.safetensors",
    "model.layers.61.mlp.shared_expert_gate.weight": "model-00141-of-00213.safetensors",
    "model.layers.61.post_attention_layernorm.weight": "model-00141-of-00213.safetensors",
    "model.layers.62.input_layernorm.weight": "model-00141-of-00213.safetensors",
    "model.layers.62.linear_attn.A_log": "model-00141-of-00213.safetensors",
    "model.layers.62.linear_attn.conv1d.weight": "model-00141-of-00213.safetensors",
    "model.layers.62.linear_attn.dt_bias": "model-00141-of-00213.safetensors",
    "model.layers.62.linear_attn.in_proj_a.weight": "model-00141-of-00213.safetensors",
    "model.layers.62.linear_attn.in_proj_b.weight": "model-00141-of-00213.safetensors",
    "model.layers.62.linear_attn.in_proj_qkv.weight": "model-00150-of-00213.safetensors",
    "model.layers.62.linear_attn.in_proj_z.weight": "model-00150-of-00213.safetensors",
    "model.layers.62.linear_attn.norm.weight": "model-00150-of-00213.safetensors",
    "model.layers.62.linear_attn.out_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.62.mlp.experts.down_proj": "model-00143-of-00213.safetensors",
    "model.layers.62.mlp.experts.gate_up_proj": "model-00142-of-00213.safetensors",
    "model.layers.62.mlp.gate.weight": "model-00150-of-00213.safetensors",
    "model.layers.62.mlp.shared_expert.down_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.62.mlp.shared_expert.gate_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.62.mlp.shared_expert.up_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.62.mlp.shared_expert_gate.weight": "model-00150-of-00213.safetensors",
    "model.layers.62.post_attention_layernorm.weight": "model-00150-of-00213.safetensors",
    "model.layers.63.input_layernorm.weight": "model-00150-of-00213.safetensors",
    "model.layers.63.mlp.experts.down_proj": "model-00145-of-00213.safetensors",
    "model.layers.63.mlp.experts.gate_up_proj": "model-00144-of-00213.safetensors",
    "model.layers.63.mlp.gate.weight": "model-00150-of-00213.safetensors",
    "model.layers.63.mlp.shared_expert.down_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.63.mlp.shared_expert.gate_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.63.mlp.shared_expert.up_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.63.mlp.shared_expert_gate.weight": "model-00150-of-00213.safetensors",
    "model.layers.63.post_attention_layernorm.weight": "model-00150-of-00213.safetensors",
    "model.layers.63.self_attn.k_norm.weight": "model-00150-of-00213.safetensors",
    "model.layers.63.self_attn.k_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.63.self_attn.o_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.63.self_attn.q_norm.weight": "model-00150-of-00213.safetensors",
    "model.layers.63.self_attn.q_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.63.self_attn.v_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.64.input_layernorm.weight": "model-00150-of-00213.safetensors",
    "model.layers.64.linear_attn.A_log": "model-00150-of-00213.safetensors",
    "model.layers.64.linear_attn.conv1d.weight": "model-00150-of-00213.safetensors",
    "model.layers.64.linear_attn.dt_bias": "model-00150-of-00213.safetensors",
    "model.layers.64.linear_attn.in_proj_a.weight": "model-00150-of-00213.safetensors",
    "model.layers.64.linear_attn.in_proj_b.weight": "model-00150-of-00213.safetensors",
    "model.layers.64.linear_attn.in_proj_qkv.weight": "model-00150-of-00213.safetensors",
    "model.layers.64.linear_attn.in_proj_z.weight": "model-00150-of-00213.safetensors",
    "model.layers.64.linear_attn.norm.weight": "model-00150-of-00213.safetensors",
    "model.layers.64.linear_attn.out_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.64.mlp.experts.down_proj": "model-00147-of-00213.safetensors",
    "model.layers.64.mlp.experts.gate_up_proj": "model-00146-of-00213.safetensors",
    "model.layers.64.mlp.gate.weight": "model-00150-of-00213.safetensors",
    "model.layers.64.mlp.shared_expert.down_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.64.mlp.shared_expert.gate_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.64.mlp.shared_expert.up_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.64.mlp.shared_expert_gate.weight": "model-00150-of-00213.safetensors",
    "model.layers.64.post_attention_layernorm.weight": "model-00150-of-00213.safetensors",
    "model.layers.65.input_layernorm.weight": "model-00150-of-00213.safetensors",
    "model.layers.65.linear_attn.A_log": "model-00150-of-00213.safetensors",
    "model.layers.65.linear_attn.conv1d.weight": "model-00150-of-00213.safetensors",
    "model.layers.65.linear_attn.dt_bias": "model-00150-of-00213.safetensors",
    "model.layers.65.linear_attn.in_proj_a.weight": "model-00150-of-00213.safetensors",
    "model.layers.65.linear_attn.in_proj_b.weight": "model-00150-of-00213.safetensors",
    "model.layers.65.linear_attn.in_proj_qkv.weight": "model-00150-of-00213.safetensors",
    "model.layers.65.linear_attn.in_proj_z.weight": "model-00150-of-00213.safetensors",
    "model.layers.65.linear_attn.norm.weight": "model-00150-of-00213.safetensors",
    "model.layers.65.linear_attn.out_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.65.mlp.experts.down_proj": "model-00149-of-00213.safetensors",
    "model.layers.65.mlp.experts.gate_up_proj": "model-00148-of-00213.safetensors",
    "model.layers.65.mlp.gate.weight": "model-00150-of-00213.safetensors",
    "model.layers.65.mlp.shared_expert.down_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.65.mlp.shared_expert.gate_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.65.mlp.shared_expert.up_proj.weight": "model-00150-of-00213.safetensors",
    "model.layers.65.mlp.shared_expert_gate.weight": "model-00150-of-00213.safetensors",
    "model.layers.65.post_attention_layernorm.weight": "model-00150-of-00213.safetensors",
    "model.layers.66.input_layernorm.weight": "model-00150-of-00213.safetensors",
    "model.layers.66.linear_attn.A_log": "model-00150-of-00213.safetensors",
    "model.layers.66.linear_attn.conv1d.weight": "model-00150-of-00213.safetensors",
    "model.layers.66.linear_attn.dt_bias": "model-00150-of-00213.safetensors",
    "model.layers.66.linear_attn.in_proj_a.weight": "model-00150-of-00213.safetensors",
    "model.layers.66.linear_attn.in_proj_b.weight": "model-00150-of-00213.safetensors",
    "model.layers.66.linear_attn.in_proj_qkv.weight": "model-00159-of-00213.safetensors",
    "model.layers.66.linear_attn.in_proj_z.weight": "model-00159-of-00213.safetensors",
    "model.layers.66.linear_attn.norm.weight": "model-00159-of-00213.safetensors",
    "model.layers.66.linear_attn.out_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.66.mlp.experts.down_proj": "model-00152-of-00213.safetensors",
    "model.layers.66.mlp.experts.gate_up_proj": "model-00151-of-00213.safetensors",
    "model.layers.66.mlp.gate.weight": "model-00159-of-00213.safetensors",
    "model.layers.66.mlp.shared_expert.down_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.66.mlp.shared_expert.gate_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.66.mlp.shared_expert.up_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.66.mlp.shared_expert_gate.weight": "model-00159-of-00213.safetensors",
    "model.layers.66.post_attention_layernorm.weight": "model-00159-of-00213.safetensors",
    "model.layers.67.input_layernorm.weight": "model-00159-of-00213.safetensors",
    "model.layers.67.mlp.experts.down_proj": "model-00154-of-00213.safetensors",
    "model.layers.67.mlp.experts.gate_up_proj": "model-00153-of-00213.safetensors",
    "model.layers.67.mlp.gate.weight": "model-00159-of-00213.safetensors",
    "model.layers.67.mlp.shared_expert.down_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.67.mlp.shared_expert.gate_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.67.mlp.shared_expert.up_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.67.mlp.shared_expert_gate.weight": "model-00159-of-00213.safetensors",
    "model.layers.67.post_attention_layernorm.weight": "model-00159-of-00213.safetensors",
    "model.layers.67.self_attn.k_norm.weight": "model-00159-of-00213.safetensors",
    "model.layers.67.self_attn.k_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.67.self_attn.o_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.67.self_attn.q_norm.weight": "model-00159-of-00213.safetensors",
    "model.layers.67.self_attn.q_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.67.self_attn.v_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.68.input_layernorm.weight": "model-00159-of-00213.safetensors",
    "model.layers.68.linear_attn.A_log": "model-00159-of-00213.safetensors",
    "model.layers.68.linear_attn.conv1d.weight": "model-00159-of-00213.safetensors",
    "model.layers.68.linear_attn.dt_bias": "model-00159-of-00213.safetensors",
    "model.layers.68.linear_attn.in_proj_a.weight": "model-00159-of-00213.safetensors",
    "model.layers.68.linear_attn.in_proj_b.weight": "model-00159-of-00213.safetensors",
    "model.layers.68.linear_attn.in_proj_qkv.weight": "model-00159-of-00213.safetensors",
    "model.layers.68.linear_attn.in_proj_z.weight": "model-00159-of-00213.safetensors",
    "model.layers.68.linear_attn.norm.weight": "model-00159-of-00213.safetensors",
    "model.layers.68.linear_attn.out_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.68.mlp.experts.down_proj": "model-00156-of-00213.safetensors",
    "model.layers.68.mlp.experts.gate_up_proj": "model-00155-of-00213.safetensors",
    "model.layers.68.mlp.gate.weight": "model-00159-of-00213.safetensors",
    "model.layers.68.mlp.shared_expert.down_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.68.mlp.shared_expert.gate_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.68.mlp.shared_expert.up_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.68.mlp.shared_expert_gate.weight": "model-00159-of-00213.safetensors",
    "model.layers.68.post_attention_layernorm.weight": "model-00159-of-00213.safetensors",
    "model.layers.69.input_layernorm.weight": "model-00159-of-00213.safetensors",
    "model.layers.69.linear_attn.A_log": "model-00159-of-00213.safetensors",
    "model.layers.69.linear_attn.conv1d.weight": "model-00159-of-00213.safetensors",
    "model.layers.69.linear_attn.dt_bias": "model-00159-of-00213.safetensors",
    "model.layers.69.linear_attn.in_proj_a.weight": "model-00159-of-00213.safetensors",
    "model.layers.69.linear_attn.in_proj_b.weight": "model-00159-of-00213.safetensors",
    "model.layers.69.linear_attn.in_proj_qkv.weight": "model-00159-of-00213.safetensors",
    "model.layers.69.linear_attn.in_proj_z.weight": "model-00159-of-00213.safetensors",
    "model.layers.69.linear_attn.norm.weight": "model-00159-of-00213.safetensors",
    "model.layers.69.linear_attn.out_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.69.mlp.experts.down_proj": "model-00158-of-00213.safetensors",
    "model.layers.69.mlp.experts.gate_up_proj": "model-00157-of-00213.safetensors",
    "model.layers.69.mlp.gate.weight": "model-00159-of-00213.safetensors",
    "model.layers.69.mlp.shared_expert.down_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.69.mlp.shared_expert.gate_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.69.mlp.shared_expert.up_proj.weight": "model-00159-of-00213.safetensors",
    "model.layers.69.mlp.shared_expert_gate.weight": "model-00159-of-00213.safetensors",
    "model.layers.69.post_attention_layernorm.weight": "model-00159-of-00213.safetensors",
    "model.layers.7.input_layernorm.weight": "model-00026-of-00213.safetensors",
    "model.layers.7.mlp.experts.down_proj": "model-00019-of-00213.safetensors",
    "model.layers.7.mlp.experts.gate_up_proj": "model-00018-of-00213.safetensors",
    "model.layers.7.mlp.gate.weight": "model-00020-of-00213.safetensors",
    "model.layers.7.mlp.shared_expert.down_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.7.mlp.shared_expert.gate_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.7.mlp.shared_expert.up_proj.weight": "model-00020-of-00213.safetensors",
    "model.layers.7.mlp.shared_expert_gate.weight": "model-00020-of-00213.safetensors",
    "model.layers.7.post_attention_layernorm.weight": "model-00020-of-00213.safetensors",
    "model.layers.7.self_attn.k_norm.weight": "model-00020-of-00213.safetensors",
    "model.layers.7.self_attn.k_proj.weight": "model-00026-of-00213.safetensors",
    "model.layers.7.self_attn.o_proj.weight": "model-00026-of-00213.safetensors",
    "model.layers.7.self_attn.q_norm.weight": "model-00026-of-00213.safetensors",
    "model.layers.7.self_attn.q_proj.weight": "model-00026-of-00213.safetensors",
    "model.layers.7.self_attn.v_proj.weight": "model-00026-of-00213.safetensors",
    "model.layers.70.input_layernorm.weight": "model-00168-of-00213.safetensors",
    "model.layers.70.linear_attn.A_log": "model-00168-of-00213.safetensors",
    "model.layers.70.linear_attn.conv1d.weight": "model-00168-of-00213.safetensors",
    "model.layers.70.linear_attn.dt_bias": "model-00168-of-00213.safetensors",
    "model.layers.70.linear_attn.in_proj_a.weight": "model-00168-of-00213.safetensors",
    "model.layers.70.linear_attn.in_proj_b.weight": "model-00168-of-00213.safetensors",
    "model.layers.70.linear_attn.in_proj_qkv.weight": "model-00168-of-00213.safetensors",
    "model.layers.70.linear_attn.in_proj_z.weight": "model-00168-of-00213.safetensors",
    "model.layers.70.linear_attn.norm.weight": "model-00168-of-00213.safetensors",
    "model.layers.70.linear_attn.out_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.70.mlp.experts.down_proj": "model-00161-of-00213.safetensors",
    "model.layers.70.mlp.experts.gate_up_proj": "model-00160-of-00213.safetensors",
    "model.layers.70.mlp.gate.weight": "model-00168-of-00213.safetensors",
    "model.layers.70.mlp.shared_expert.down_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.70.mlp.shared_expert.gate_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.70.mlp.shared_expert.up_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.70.mlp.shared_expert_gate.weight": "model-00168-of-00213.safetensors",
    "model.layers.70.post_attention_layernorm.weight": "model-00168-of-00213.safetensors",
    "model.layers.71.input_layernorm.weight": "model-00168-of-00213.safetensors",
    "model.layers.71.mlp.experts.down_proj": "model-00163-of-00213.safetensors",
    "model.layers.71.mlp.experts.gate_up_proj": "model-00162-of-00213.safetensors",
    "model.layers.71.mlp.gate.weight": "model-00168-of-00213.safetensors",
    "model.layers.71.mlp.shared_expert.down_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.71.mlp.shared_expert.gate_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.71.mlp.shared_expert.up_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.71.mlp.shared_expert_gate.weight": "model-00168-of-00213.safetensors",
    "model.layers.71.post_attention_layernorm.weight": "model-00168-of-00213.safetensors",
    "model.layers.71.self_attn.k_norm.weight": "model-00168-of-00213.safetensors",
    "model.layers.71.self_attn.k_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.71.self_attn.o_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.71.self_attn.q_norm.weight": "model-00168-of-00213.safetensors",
    "model.layers.71.self_attn.q_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.71.self_attn.v_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.72.input_layernorm.weight": "model-00168-of-00213.safetensors",
    "model.layers.72.linear_attn.A_log": "model-00168-of-00213.safetensors",
    "model.layers.72.linear_attn.conv1d.weight": "model-00168-of-00213.safetensors",
    "model.layers.72.linear_attn.dt_bias": "model-00168-of-00213.safetensors",
    "model.layers.72.linear_attn.in_proj_a.weight": "model-00168-of-00213.safetensors",
    "model.layers.72.linear_attn.in_proj_b.weight": "model-00168-of-00213.safetensors",
    "model.layers.72.linear_attn.in_proj_qkv.weight": "model-00168-of-00213.safetensors",
    "model.layers.72.linear_attn.in_proj_z.weight": "model-00168-of-00213.safetensors",
    "model.layers.72.linear_attn.norm.weight": "model-00168-of-00213.safetensors",
    "model.layers.72.linear_attn.out_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.72.mlp.experts.down_proj": "model-00165-of-00213.safetensors",
    "model.layers.72.mlp.experts.gate_up_proj": "model-00164-of-00213.safetensors",
    "model.layers.72.mlp.gate.weight": "model-00168-of-00213.safetensors",
    "model.layers.72.mlp.shared_expert.down_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.72.mlp.shared_expert.gate_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.72.mlp.shared_expert.up_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.72.mlp.shared_expert_gate.weight": "model-00168-of-00213.safetensors",
    "model.layers.72.post_attention_layernorm.weight": "model-00168-of-00213.safetensors",
    "model.layers.73.input_layernorm.weight": "model-00168-of-00213.safetensors",
    "model.layers.73.linear_attn.A_log": "model-00168-of-00213.safetensors",
    "model.layers.73.linear_attn.conv1d.weight": "model-00168-of-00213.safetensors",
    "model.layers.73.linear_attn.dt_bias": "model-00168-of-00213.safetensors",
    "model.layers.73.linear_attn.in_proj_a.weight": "model-00168-of-00213.safetensors",
    "model.layers.73.linear_attn.in_proj_b.weight": "model-00168-of-00213.safetensors",
    "model.layers.73.linear_attn.in_proj_qkv.weight": "model-00168-of-00213.safetensors",
    "model.layers.73.linear_attn.in_proj_z.weight": "model-00168-of-00213.safetensors",
    "model.layers.73.linear_attn.norm.weight": "model-00168-of-00213.safetensors",
    "model.layers.73.linear_attn.out_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.73.mlp.experts.down_proj": "model-00167-of-00213.safetensors",
    "model.layers.73.mlp.experts.gate_up_proj": "model-00166-of-00213.safetensors",
    "model.layers.73.mlp.gate.weight": "model-00168-of-00213.safetensors",
    "model.layers.73.mlp.shared_expert.down_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.73.mlp.shared_expert.gate_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.73.mlp.shared_expert.up_proj.weight": "model-00168-of-00213.safetensors",
    "model.layers.73.mlp.shared_expert_gate.weight": "model-00168-of-00213.safetensors",
    "model.layers.73.post_attention_layernorm.weight": "model-00168-of-00213.safetensors",
    "model.layers.74.input_layernorm.weight": "model-00168-of-00213.safetensors",
    "model.layers.74.linear_attn.A_log": "model-00168-of-00213.safetensors",
    "model.layers.74.linear_attn.conv1d.weight": "model-00168-of-00213.safetensors",
    "model.layers.74.linear_attn.dt_bias": "model-00168-of-00213.safetensors",
    "model.layers.74.linear_attn.in_proj_a.weight": "model-00168-of-00213.safetensors",
    "model.layers.74.linear_attn.in_proj_b.weight": "model-00168-of-00213.safetensors",
    "model.layers.74.linear_attn.in_proj_qkv.weight": "model-00177-of-00213.safetensors",
    "model.layers.74.linear_attn.in_proj_z.weight": "model-00177-of-00213.safetensors",
    "model.layers.74.linear_attn.norm.weight": "model-00177-of-00213.safetensors",
    "model.layers.74.linear_attn.out_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.74.mlp.experts.down_proj": "model-00170-of-00213.safetensors",
    "model.layers.74.mlp.experts.gate_up_proj": "model-00169-of-00213.safetensors",
    "model.layers.74.mlp.gate.weight": "model-00177-of-00213.safetensors",
    "model.layers.74.mlp.shared_expert.down_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.74.mlp.shared_expert.gate_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.74.mlp.shared_expert.up_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.74.mlp.shared_expert_gate.weight": "model-00177-of-00213.safetensors",
    "model.layers.74.post_attention_layernorm.weight": "model-00177-of-00213.safetensors",
    "model.layers.75.input_layernorm.weight": "model-00177-of-00213.safetensors",
    "model.layers.75.mlp.experts.down_proj": "model-00172-of-00213.safetensors",
    "model.layers.75.mlp.experts.gate_up_proj": "model-00171-of-00213.safetensors",
    "model.layers.75.mlp.gate.weight": "model-00177-of-00213.safetensors",
    "model.layers.75.mlp.shared_expert.down_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.75.mlp.shared_expert.gate_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.75.mlp.shared_expert.up_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.75.mlp.shared_expert_gate.weight": "model-00177-of-00213.safetensors",
    "model.layers.75.post_attention_layernorm.weight": "model-00177-of-00213.safetensors",
    "model.layers.75.self_attn.k_norm.weight": "model-00177-of-00213.safetensors",
    "model.layers.75.self_attn.k_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.75.self_attn.o_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.75.self_attn.q_norm.weight": "model-00177-of-00213.safetensors",
    "model.layers.75.self_attn.q_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.75.self_attn.v_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.76.input_layernorm.weight": "model-00177-of-00213.safetensors",
    "model.layers.76.linear_attn.A_log": "model-00177-of-00213.safetensors",
    "model.layers.76.linear_attn.conv1d.weight": "model-00177-of-00213.safetensors",
    "model.layers.76.linear_attn.dt_bias": "model-00177-of-00213.safetensors",
    "model.layers.76.linear_attn.in_proj_a.weight": "model-00177-of-00213.safetensors",
    "model.layers.76.linear_attn.in_proj_b.weight": "model-00177-of-00213.safetensors",
    "model.layers.76.linear_attn.in_proj_qkv.weight": "model-00177-of-00213.safetensors",
    "model.layers.76.linear_attn.in_proj_z.weight": "model-00177-of-00213.safetensors",
    "model.layers.76.linear_attn.norm.weight": "model-00177-of-00213.safetensors",
    "model.layers.76.linear_attn.out_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.76.mlp.experts.down_proj": "model-00174-of-00213.safetensors",
    "model.layers.76.mlp.experts.gate_up_proj": "model-00173-of-00213.safetensors",
    "model.layers.76.mlp.gate.weight": "model-00177-of-00213.safetensors",
    "model.layers.76.mlp.shared_expert.down_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.76.mlp.shared_expert.gate_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.76.mlp.shared_expert.up_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.76.mlp.shared_expert_gate.weight": "model-00177-of-00213.safetensors",
    "model.layers.76.post_attention_layernorm.weight": "model-00177-of-00213.safetensors",
    "model.layers.77.input_layernorm.weight": "model-00177-of-00213.safetensors",
    "model.layers.77.linear_attn.A_log": "model-00177-of-00213.safetensors",
    "model.layers.77.linear_attn.conv1d.weight": "model-00177-of-00213.safetensors",
    "model.layers.77.linear_attn.dt_bias": "model-00177-of-00213.safetensors",
    "model.layers.77.linear_attn.in_proj_a.weight": "model-00177-of-00213.safetensors",
    "model.layers.77.linear_attn.in_proj_b.weight": "model-00177-of-00213.safetensors",
    "model.layers.77.linear_attn.in_proj_qkv.weight": "model-00177-of-00213.safetensors",
    "model.layers.77.linear_attn.in_proj_z.weight": "model-00177-of-00213.safetensors",
    "model.layers.77.linear_attn.norm.weight": "model-00177-of-00213.safetensors",
    "model.layers.77.linear_attn.out_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.77.mlp.experts.down_proj": "model-00176-of-00213.safetensors",
    "model.layers.77.mlp.experts.gate_up_proj": "model-00175-of-00213.safetensors",
    "model.layers.77.mlp.gate.weight": "model-00177-of-00213.safetensors",
    "model.layers.77.mlp.shared_expert.down_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.77.mlp.shared_expert.gate_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.77.mlp.shared_expert.up_proj.weight": "model-00177-of-00213.safetensors",
    "model.layers.77.mlp.shared_expert_gate.weight": "model-00177-of-00213.safetensors",
    "model.layers.77.post_attention_layernorm.weight": "model-00177-of-00213.safetensors",
    "model.layers.78.input_layernorm.weight": "model-00177-of-00213.safetensors",
    "model.layers.78.linear_attn.A_log": "model-00177-of-00213.safetensors",
    "model.layers.78.linear_attn.conv1d.weight": "model-00177-of-00213.safetensors",
    "model.layers.78.linear_attn.dt_bias": "model-00177-of-00213.safetensors",
    "model.layers.78.linear_attn.in_proj_a.weight": "model-00177-of-00213.safetensors",
    "model.layers.78.linear_attn.in_proj_b.weight": "model-00177-of-00213.safetensors",
    "model.layers.78.linear_attn.in_proj_qkv.weight": "model-00184-of-00213.safetensors",
    "model.layers.78.linear_attn.in_proj_z.weight": "model-00184-of-00213.safetensors",
    "model.layers.78.linear_attn.norm.weight": "model-00184-of-00213.safetensors",
    "model.layers.78.linear_attn.out_proj.weight": "model-00184-of-00213.safetensors",
    "model.layers.78.mlp.experts.down_proj": "model-00179-of-00213.safetensors",
    "model.layers.78.mlp.experts.gate_up_proj": "model-00178-of-00213.safetensors",
    "model.layers.78.mlp.gate.weight": "model-00184-of-00213.safetensors",
    "model.layers.78.mlp.shared_expert.down_proj.weight": "model-00184-of-00213.safetensors",
    "model.layers.78.mlp.shared_expert.gate_proj.weight": "model-00184-of-00213.safetensors",
    "model.layers.78.mlp.shared_expert.up_proj.weight": "model-00184-of-00213.safetensors",
    "model.layers.78.mlp.shared_expert_gate.weight": "model-00184-of-00213.safetensors",
    "model.layers.78.post_attention_layernorm.weight": "model-00184-of-00213.safetensors",
    "model.layers.79.input_layernorm.weight": "model-00184-of-00213.safetensors",
    "model.layers.79.mlp.experts.down_proj": "model-00181-of-00213.safetensors",
    "model.layers.79.mlp.experts.gate_up_proj": "model-00180-of-00213.safetensors",
    "model.layers.79.mlp.gate.weight": "model-00184-of-00213.safetensors",
    "model.layers.79.mlp.shared_expert.down_proj.weight": "model-00184-of-00213.safetensors",
    "model.layers.79.mlp.shared_expert.gate_proj.weight": "model-00184-of-00213.safetensors",
    "model.layers.79.mlp.shared_expert.up_proj.weight": "model-00184-of-00213.safetensors",
    "model.layers.79.mlp.shared_expert_gate.weight": "model-00184-of-00213.safetensors",
    "model.layers.79.post_attention_layernorm.weight": "model-00184-of-00213.safetensors",
    "model.layers.79.self_attn.k_norm.weight": "model-00184-of-00213.safetensors",
    "model.layers.79.self_attn.k_proj.weight": "model-00184-of-00213.safetensors",
    "model.layers.79.self_attn.o_proj.weight": "model-00184-of-00213.safetensors",
    "model.layers.79.self_attn.q_norm.weight": "model-00184-of-00213.safetensors",
    "model.layers.79.self_attn.q_proj.weight": "model-00184-of-00213.safetensors",
    "model.layers.79.self_attn.v_proj.weight": "model-00184-of-00213.safetensors",
    "model.layers.8.input_layernorm.weight": "model-00026-of-00213.safetensors",
    "model.layers.8.linear_attn.A_log": "model-00026-of-00213.safetensors",
    "model.layers.8.linear_attn.conv1d.weight": "model-00026-of-00213.safetensors",
    "model.layers.8.linear_attn.dt_bias": "model-00026-of-00213.safetensors",
    "model.layers.8.linear_attn.in_proj_a.weight": "model-00026-of-00213.safetensors",
    "model.layers.8.linear_attn.in_proj_b.weight": "model-00026-of-00213.safetensors",
    "model.layers.8.linear_attn.in_proj_qkv.weight": "model-00026-of-00213.safetensors",
    "model.layers.8.linear_attn.in_proj_z.weight": "model-00026-of-00213.safetensors",
    "model.layers.8.linear_attn.norm.weight": "model-00026-of-00213.safetensors",
    "model.layers.8.linear_attn.out_proj.weight": "model-00026-of-00213.safetensors",
    "model.layers.8.mlp.experts.down_proj": "model-00022-of-00213.safetensors",
    "model.layers.8.mlp.experts.gate_up_proj": "model-00021-of-00213.safetensors",
    "model.layers.8.mlp.gate.weight": "model-00026-of-00213.safetensors",
    "model.layers.8.mlp.shared_expert.down_proj.weight": "model-00026-of-00213.safetensors",
    "model.layers.8.mlp.shared_expert.gate_proj.weight": "model-00026-of-00213.safetensors",
    "model.layers.8.mlp.shared_expert.up_proj.weight": "model-00026-of-00213.safetensors",
    "model.layers.8.mlp.shared_expert_gate.weight": "model-00026-of-00213.safetensors",
    "model.layers.8.post_attention_layernorm.weight": "model-00026-of-00213.safetensors",
    "model.layers.80.input_layernorm.weight": "model-00184-of-00213.safetensors",
    "model.layers.80.linear_attn.A_log": "model-00184-of-00213.safetensors",
    "model.layers.80.linear_attn.conv1d.weight": "model-00184-of-00213.safetensors",
    "model.layers.80.linear_attn.dt_bias": "model-00184-of-00213.safetensors",
    "model.layers.80.linear_attn.in_proj_a.weight": "model-00184-of-00213.safetensors",
    "model.layers.80.linear_attn.in_proj_b.weight": "model-00184-of-00213.safetensors",
    "model.layers.80.linear_attn.in_proj_qkv.weight": "model-00184-of-00213.safetensors",
    "model.layers.80.linear_attn.in_proj_z.weight": "model-00184-of-00213.safetensors",
    "model.layers.80.linear_attn.norm.weight": "model-00184-of-00213.safetensors",
    "model.layers.80.linear_attn.out_proj.weight": "model-00184-of-00213.safetensors",
    "model.layers.80.mlp.experts.down_proj": "model-00183-of-00213.safetensors",
    "model.layers.80.mlp.experts.gate_up_proj": "model-00182-of-00213.safetensors",
    "model.layers.80.mlp.gate.weight": "model-00184-of-00213.safetensors",
    "model.layers.80.mlp.shared_expert.down_proj.weight": "model-00184-of-00213.safetensors",
    "model.layers.80.mlp.shared_expert.gate_proj.weight": "model-00184-of-00213.safetensors",
    "model.layers.80.mlp.shared_expert.up_proj.weight": "model-00184-of-00213.safetensors",
    "model.layers.80.mlp.shared_expert_gate.weight": "model-00184-of-00213.safetensors",
    "model.layers.80.post_attention_layernorm.weight": "model-00184-of-00213.safetensors",
    "model.layers.81.input_layernorm.weight": "model-00193-of-00213.safetensors",
    "model.layers.81.linear_attn.A_log": "model-00193-of-00213.safetensors",
    "model.layers.81.linear_attn.conv1d.weight": "model-00193-of-00213.safetensors",
    "model.layers.81.linear_attn.dt_bias": "model-00193-of-00213.safetensors",
    "model.layers.81.linear_attn.in_proj_a.weight": "model-00193-of-00213.safetensors",
    "model.layers.81.linear_attn.in_proj_b.weight": "model-00193-of-00213.safetensors",
    "model.layers.81.linear_attn.in_proj_qkv.weight": "model-00193-of-00213.safetensors",
    "model.layers.81.linear_attn.in_proj_z.weight": "model-00193-of-00213.safetensors",
    "model.layers.81.linear_attn.norm.weight": "model-00193-of-00213.safetensors",
    "model.layers.81.linear_attn.out_proj.weight": "model-00193-of-00213.safetensors",
    "model.layers.81.mlp.experts.down_proj": "model-00188-of-00213.safetensors",
    "model.layers.81.mlp.experts.gate_up_proj": "model-00187-of-00213.safetensors",
    "model.layers.81.mlp.gate.weight": "model-00193-of-00213.safetensors",
    "model.layers.81.mlp.shared_expert.down_proj.weight": "model-00193-of-00213.safetensors",
    "model.layers.81.mlp.shared_expert.gate_proj.weight": "model-00193-of-00213.safetensors",
    "model.layers.81.mlp.shared_expert.up_proj.weight": "model-00193-of-00213.safetensors",
    "model.layers.81.mlp.shared_expert_gate.weight": "model-00193-of-00213.safetensors",
    "model.layers.81.post_attention_layernorm.weight": "model-00193-of-00213.safetensors",
    "model.layers.82.input_layernorm.weight": "model-00193-of-00213.safetensors",
    "model.layers.82.linear_attn.A_log": "model-00193-of-00213.safetensors",
    "model.layers.82.linear_attn.conv1d.weight": "model-00193-of-00213.safetensors",
    "model.layers.82.linear_attn.dt_bias": "model-00193-of-00213.safetensors",
    "model.layers.82.linear_attn.in_proj_a.weight": "model-00193-of-00213.safetensors",
    "model.layers.82.linear_attn.in_proj_b.weight": "model-00193-of-00213.safetensors",
    "model.layers.82.linear_attn.in_proj_qkv.weight": "model-00193-of-00213.safetensors",
    "model.layers.82.linear_attn.in_proj_z.weight": "model-00193-of-00213.safetensors",
    "model.layers.82.linear_attn.norm.weight": "model-00193-of-00213.safetensors",
    "model.layers.82.linear_attn.out_proj.weight": "model-00193-of-00213.safetensors",
    "model.layers.82.mlp.experts.down_proj": "model-00190-of-00213.safetensors",
    "model.layers.82.mlp.experts.gate_up_proj": "model-00189-of-00213.safetensors",
    "model.layers.82.mlp.gate.weight": "model-00193-of-00213.safetensors",
    "model.layers.82.mlp.shared_expert.down_proj.weight": "model-00193-of-00213.safetensors",
    "model.layers.82.mlp.shared_expert.gate_proj.weight": "model-00193-of-00213.safetensors",
    "model.layers.82.mlp.shared_expert.up_proj.weight": "model-00193-of-00213.safetensors",
    "model.layers.82.mlp.shared_expert_gate.weight": "model-00193-of-00213.safetensors",
    "model.layers.82.post_attention_layernorm.weight": "model-00193-of-00213.safetensors",
    "model.layers.83.input_layernorm.weight": "model-00193-of-00213.safetensors",
    "model.layers.83.mlp.experts.down_proj": "model-00192-of-00213.safetensors",
    "model.layers.83.mlp.experts.gate_up_proj": "model-00191-of-00213.safetensors",
    "model.layers.83.mlp.gate.weight": "model-00193-of-00213.safetensors",
    "model.layers.83.mlp.shared_expert.down_proj.weight": "model-00193-of-00213.safetensors",
    "model.layers.83.mlp.shared_expert.gate_proj.weight": "model-00193-of-00213.safetensors",
    "model.layers.83.mlp.shared_expert.up_proj.weight": "model-00193-of-00213.safetensors",
    "model.layers.83.mlp.shared_expert_gate.weight": "model-00193-of-00213.safetensors",
    "model.layers.83.post_attention_layernorm.weight": "model-00193-of-00213.safetensors",
    "model.layers.83.self_attn.k_norm.weight": "model-00193-of-00213.safetensors",
    "model.layers.83.self_attn.k_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.83.self_attn.o_proj.weight": "model-00193-of-00213.safetensors",
    "model.layers.83.self_attn.q_norm.weight": "model-00202-of-00213.safetensors",
    "model.layers.83.self_attn.q_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.83.self_attn.v_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.84.input_layernorm.weight": "model-00202-of-00213.safetensors",
    "model.layers.84.linear_attn.A_log": "model-00202-of-00213.safetensors",
    "model.layers.84.linear_attn.conv1d.weight": "model-00202-of-00213.safetensors",
    "model.layers.84.linear_attn.dt_bias": "model-00202-of-00213.safetensors",
    "model.layers.84.linear_attn.in_proj_a.weight": "model-00202-of-00213.safetensors",
    "model.layers.84.linear_attn.in_proj_b.weight": "model-00202-of-00213.safetensors",
    "model.layers.84.linear_attn.in_proj_qkv.weight": "model-00202-of-00213.safetensors",
    "model.layers.84.linear_attn.in_proj_z.weight": "model-00202-of-00213.safetensors",
    "model.layers.84.linear_attn.norm.weight": "model-00202-of-00213.safetensors",
    "model.layers.84.linear_attn.out_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.84.mlp.experts.down_proj": "model-00195-of-00213.safetensors",
    "model.layers.84.mlp.experts.gate_up_proj": "model-00194-of-00213.safetensors",
    "model.layers.84.mlp.gate.weight": "model-00202-of-00213.safetensors",
    "model.layers.84.mlp.shared_expert.down_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.84.mlp.shared_expert.gate_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.84.mlp.shared_expert.up_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.84.mlp.shared_expert_gate.weight": "model-00202-of-00213.safetensors",
    "model.layers.84.post_attention_layernorm.weight": "model-00202-of-00213.safetensors",
    "model.layers.85.input_layernorm.weight": "model-00202-of-00213.safetensors",
    "model.layers.85.linear_attn.A_log": "model-00202-of-00213.safetensors",
    "model.layers.85.linear_attn.conv1d.weight": "model-00202-of-00213.safetensors",
    "model.layers.85.linear_attn.dt_bias": "model-00202-of-00213.safetensors",
    "model.layers.85.linear_attn.in_proj_a.weight": "model-00202-of-00213.safetensors",
    "model.layers.85.linear_attn.in_proj_b.weight": "model-00202-of-00213.safetensors",
    "model.layers.85.linear_attn.in_proj_qkv.weight": "model-00202-of-00213.safetensors",
    "model.layers.85.linear_attn.in_proj_z.weight": "model-00202-of-00213.safetensors",
    "model.layers.85.linear_attn.norm.weight": "model-00202-of-00213.safetensors",
    "model.layers.85.linear_attn.out_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.85.mlp.experts.down_proj": "model-00197-of-00213.safetensors",
    "model.layers.85.mlp.experts.gate_up_proj": "model-00196-of-00213.safetensors",
    "model.layers.85.mlp.gate.weight": "model-00202-of-00213.safetensors",
    "model.layers.85.mlp.shared_expert.down_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.85.mlp.shared_expert.gate_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.85.mlp.shared_expert.up_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.85.mlp.shared_expert_gate.weight": "model-00202-of-00213.safetensors",
    "model.layers.85.post_attention_layernorm.weight": "model-00202-of-00213.safetensors",
    "model.layers.86.input_layernorm.weight": "model-00202-of-00213.safetensors",
    "model.layers.86.linear_attn.A_log": "model-00202-of-00213.safetensors",
    "model.layers.86.linear_attn.conv1d.weight": "model-00202-of-00213.safetensors",
    "model.layers.86.linear_attn.dt_bias": "model-00202-of-00213.safetensors",
    "model.layers.86.linear_attn.in_proj_a.weight": "model-00202-of-00213.safetensors",
    "model.layers.86.linear_attn.in_proj_b.weight": "model-00202-of-00213.safetensors",
    "model.layers.86.linear_attn.in_proj_qkv.weight": "model-00202-of-00213.safetensors",
    "model.layers.86.linear_attn.in_proj_z.weight": "model-00202-of-00213.safetensors",
    "model.layers.86.linear_attn.norm.weight": "model-00202-of-00213.safetensors",
    "model.layers.86.linear_attn.out_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.86.mlp.experts.down_proj": "model-00199-of-00213.safetensors",
    "model.layers.86.mlp.experts.gate_up_proj": "model-00198-of-00213.safetensors",
    "model.layers.86.mlp.gate.weight": "model-00202-of-00213.safetensors",
    "model.layers.86.mlp.shared_expert.down_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.86.mlp.shared_expert.gate_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.86.mlp.shared_expert.up_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.86.mlp.shared_expert_gate.weight": "model-00202-of-00213.safetensors",
    "model.layers.86.post_attention_layernorm.weight": "model-00202-of-00213.safetensors",
    "model.layers.87.input_layernorm.weight": "model-00202-of-00213.safetensors",
    "model.layers.87.mlp.experts.down_proj": "model-00201-of-00213.safetensors",
    "model.layers.87.mlp.experts.gate_up_proj": "model-00200-of-00213.safetensors",
    "model.layers.87.mlp.gate.weight": "model-00202-of-00213.safetensors",
    "model.layers.87.mlp.shared_expert.down_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.87.mlp.shared_expert.gate_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.87.mlp.shared_expert.up_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.87.mlp.shared_expert_gate.weight": "model-00202-of-00213.safetensors",
    "model.layers.87.post_attention_layernorm.weight": "model-00202-of-00213.safetensors",
    "model.layers.87.self_attn.k_norm.weight": "model-00202-of-00213.safetensors",
    "model.layers.87.self_attn.k_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.87.self_attn.o_proj.weight": "model-00202-of-00213.safetensors",
    "model.layers.87.self_attn.q_norm.weight": "model-00211-of-00213.safetensors",
    "model.layers.87.self_attn.q_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.87.self_attn.v_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.88.input_layernorm.weight": "model-00211-of-00213.safetensors",
    "model.layers.88.linear_attn.A_log": "model-00211-of-00213.safetensors",
    "model.layers.88.linear_attn.conv1d.weight": "model-00211-of-00213.safetensors",
    "model.layers.88.linear_attn.dt_bias": "model-00211-of-00213.safetensors",
    "model.layers.88.linear_attn.in_proj_a.weight": "model-00211-of-00213.safetensors",
    "model.layers.88.linear_attn.in_proj_b.weight": "model-00211-of-00213.safetensors",
    "model.layers.88.linear_attn.in_proj_qkv.weight": "model-00211-of-00213.safetensors",
    "model.layers.88.linear_attn.in_proj_z.weight": "model-00211-of-00213.safetensors",
    "model.layers.88.linear_attn.norm.weight": "model-00211-of-00213.safetensors",
    "model.layers.88.linear_attn.out_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.88.mlp.experts.down_proj": "model-00204-of-00213.safetensors",
    "model.layers.88.mlp.experts.gate_up_proj": "model-00203-of-00213.safetensors",
    "model.layers.88.mlp.gate.weight": "model-00211-of-00213.safetensors",
    "model.layers.88.mlp.shared_expert.down_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.88.mlp.shared_expert.gate_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.88.mlp.shared_expert.up_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.88.mlp.shared_expert_gate.weight": "model-00211-of-00213.safetensors",
    "model.layers.88.post_attention_layernorm.weight": "model-00211-of-00213.safetensors",
    "model.layers.89.input_layernorm.weight": "model-00211-of-00213.safetensors",
    "model.layers.89.linear_attn.A_log": "model-00211-of-00213.safetensors",
    "model.layers.89.linear_attn.conv1d.weight": "model-00211-of-00213.safetensors",
    "model.layers.89.linear_attn.dt_bias": "model-00211-of-00213.safetensors",
    "model.layers.89.linear_attn.in_proj_a.weight": "model-00211-of-00213.safetensors",
    "model.layers.89.linear_attn.in_proj_b.weight": "model-00211-of-00213.safetensors",
    "model.layers.89.linear_attn.in_proj_qkv.weight": "model-00211-of-00213.safetensors",
    "model.layers.89.linear_attn.in_proj_z.weight": "model-00211-of-00213.safetensors",
    "model.layers.89.linear_attn.norm.weight": "model-00211-of-00213.safetensors",
    "model.layers.89.linear_attn.out_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.89.mlp.experts.down_proj": "model-00206-of-00213.safetensors",
    "model.layers.89.mlp.experts.gate_up_proj": "model-00205-of-00213.safetensors",
    "model.layers.89.mlp.gate.weight": "model-00211-of-00213.safetensors",
    "model.layers.89.mlp.shared_expert.down_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.89.mlp.shared_expert.gate_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.89.mlp.shared_expert.up_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.89.mlp.shared_expert_gate.weight": "model-00211-of-00213.safetensors",
    "model.layers.89.post_attention_layernorm.weight": "model-00211-of-00213.safetensors",
    "model.layers.9.input_layernorm.weight": "model-00026-of-00213.safetensors",
    "model.layers.9.linear_attn.A_log": "model-00026-of-00213.safetensors",
    "model.layers.9.linear_attn.conv1d.weight": "model-00026-of-00213.safetensors",
    "model.layers.9.linear_attn.dt_bias": "model-00026-of-00213.safetensors",
    "model.layers.9.linear_attn.in_proj_a.weight": "model-00026-of-00213.safetensors",
    "model.layers.9.linear_attn.in_proj_b.weight": "model-00026-of-00213.safetensors",
    "model.layers.9.linear_attn.in_proj_qkv.weight": "model-00026-of-00213.safetensors",
    "model.layers.9.linear_attn.in_proj_z.weight": "model-00026-of-00213.safetensors",
    "model.layers.9.linear_attn.norm.weight": "model-00026-of-00213.safetensors",
    "model.layers.9.linear_attn.out_proj.weight": "model-00026-of-00213.safetensors",
    "model.layers.9.mlp.experts.down_proj": "model-00024-of-00213.safetensors",
    "model.layers.9.mlp.experts.gate_up_proj": "model-00023-of-00213.safetensors",
    "model.layers.9.mlp.gate.weight": "model-00026-of-00213.safetensors",
    "model.layers.9.mlp.shared_expert.down_proj.weight": "model-00026-of-00213.safetensors",
    "model.layers.9.mlp.shared_expert.gate_proj.weight": "model-00026-of-00213.safetensors",
    "model.layers.9.mlp.shared_expert.up_proj.weight": "model-00026-of-00213.safetensors",
    "model.layers.9.mlp.shared_expert_gate.weight": "model-00026-of-00213.safetensors",
    "model.layers.9.post_attention_layernorm.weight": "model-00026-of-00213.safetensors",
    "model.layers.90.input_layernorm.weight": "model-00211-of-00213.safetensors",
    "model.layers.90.linear_attn.A_log": "model-00211-of-00213.safetensors",
    "model.layers.90.linear_attn.conv1d.weight": "model-00211-of-00213.safetensors",
    "model.layers.90.linear_attn.dt_bias": "model-00211-of-00213.safetensors",
    "model.layers.90.linear_attn.in_proj_a.weight": "model-00211-of-00213.safetensors",
    "model.layers.90.linear_attn.in_proj_b.weight": "model-00211-of-00213.safetensors",
    "model.layers.90.linear_attn.in_proj_qkv.weight": "model-00211-of-00213.safetensors",
    "model.layers.90.linear_attn.in_proj_z.weight": "model-00211-of-00213.safetensors",
    "model.layers.90.linear_attn.norm.weight": "model-00211-of-00213.safetensors",
    "model.layers.90.linear_attn.out_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.90.mlp.experts.down_proj": "model-00208-of-00213.safetensors",
    "model.layers.90.mlp.experts.gate_up_proj": "model-00207-of-00213.safetensors",
    "model.layers.90.mlp.gate.weight": "model-00211-of-00213.safetensors",
    "model.layers.90.mlp.shared_expert.down_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.90.mlp.shared_expert.gate_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.90.mlp.shared_expert.up_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.90.mlp.shared_expert_gate.weight": "model-00211-of-00213.safetensors",
    "model.layers.90.post_attention_layernorm.weight": "model-00211-of-00213.safetensors",
    "model.layers.91.input_layernorm.weight": "model-00211-of-00213.safetensors",
    "model.layers.91.mlp.experts.down_proj": "model-00210-of-00213.safetensors",
    "model.layers.91.mlp.experts.gate_up_proj": "model-00209-of-00213.safetensors",
    "model.layers.91.mlp.gate.weight": "model-00211-of-00213.safetensors",
    "model.layers.91.mlp.shared_expert.down_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.91.mlp.shared_expert.gate_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.91.mlp.shared_expert.up_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.91.mlp.shared_expert_gate.weight": "model-00211-of-00213.safetensors",
    "model.layers.91.post_attention_layernorm.weight": "model-00211-of-00213.safetensors",
    "model.layers.91.self_attn.k_norm.weight": "model-00211-of-00213.safetensors",
    "model.layers.91.self_attn.k_proj.weight": "model-00213-of-00213.safetensors",
    "model.layers.91.self_attn.o_proj.weight": "model-00211-of-00213.safetensors",
    "model.layers.91.self_attn.q_norm.weight": "model-00213-of-00213.safetensors",
    "model.layers.91.self_attn.q_proj.weight": "model-00213-of-00213.safetensors",
    "model.layers.91.self_attn.v_proj.weight": "model-00213-of-00213.safetensors",
    "model.norm.weight": "model-00193-of-00213.safetensors",
    "mtp.fc.weight": "model-00193-of-00213.safetensors",
    "mtp.layers.0.input_layernorm.weight": "model-00193-of-00213.safetensors",
    "mtp.layers.0.mlp.experts.down_proj": "model-00186-of-00213.safetensors",
    "mtp.layers.0.mlp.experts.gate_up_proj": "model-00185-of-00213.safetensors",
    "mtp.layers.0.mlp.gate.weight": "model-00193-of-00213.safetensors",
    "mtp.layers.0.mlp.shared_expert.down_proj.weight": "model-00193-of-00213.safetensors",
    "mtp.layers.0.mlp.shared_expert.gate_proj.weight": "model-00193-of-00213.safetensors",
    "mtp.layers.0.mlp.shared_expert.up_proj.weight": "model-00193-of-00213.safetensors",
    "mtp.layers.0.mlp.shared_expert_gate.weight": "model-00193-of-00213.safetensors",
    "mtp.layers.0.post_attention_layernorm.weight": "model-00193-of-00213.safetensors",
    "mtp.layers.0.self_attn.k_norm.weight": "model-00193-of-00213.safetensors",
    "mtp.layers.0.self_attn.k_proj.weight": "model-00193-of-00213.safetensors",
    "mtp.layers.0.self_attn.o_proj.weight": "model-00193-of-00213.safetensors",
    "mtp.layers.0.self_attn.q_norm.weight": "model-00193-of-00213.safetensors",
    "mtp.layers.0.self_attn.q_proj.weight": "model-00193-of-00213.safetensors",
    "mtp.layers.0.self_attn.v_proj.weight": "model-00193-of-00213.safetensors",
    "mtp.norm.weight": "model-00193-of-00213.safetensors",
    "mtp.pre_fc_norm_embedding.weight": "model-00193-of-00213.safetensors",
    "mtp.pre_fc_norm_hidden.weight": "model-00193-of-00213.safetensors"
  }
}