{ "metadata": { "total_size": 79885684736 }, "weight_map": { "lm_head.weight": "pytorch_model-00001-of-00009.bin", "model.embed_tokens.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.input_layernorm.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.0.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.0.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.1.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.1.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.10.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.10.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.11.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.11.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.12.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.12.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.13.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.13.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.14.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.14.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.15.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.15.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.16.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.16.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.17.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.17.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.18.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.18.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.19.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.19.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.2.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.2.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.20.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.20.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.21.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.21.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.22.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.22.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.23.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.23.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.24.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.24.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.25.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.25.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.26.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.26.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.27.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.27.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.28.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.28.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.29.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.29.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.3.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.3.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.30.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.30.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.31.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.31.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.4.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.4.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.5.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.5.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.6.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.6.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.7.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.7.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.8.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.8.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.9.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.experts.9.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.gate.key.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.gate.query.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.mlp.gate.value.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.self_attn.lf_gate.conv1.bias": "pytorch_model-00001-of-00009.bin", "model.layers.0.self_attn.lf_gate.conv1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.self_attn.lf_gate.conv2.bias": "pytorch_model-00001-of-00009.bin", "model.layers.0.self_attn.lf_gate.conv2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00009.bin", "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.input_layernorm.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.0.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.0.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.1.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.1.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.10.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.10.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.11.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.11.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.12.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.12.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.13.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.13.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.14.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.14.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.15.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.15.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.16.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.16.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.17.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.17.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.18.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.18.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.19.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.19.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.2.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.2.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.20.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.20.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.21.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.21.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.22.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.22.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.23.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.23.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.24.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.24.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.25.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.25.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.26.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.26.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.27.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.27.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.28.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.28.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.29.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.29.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.3.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.3.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.30.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.30.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.31.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.31.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.4.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.4.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.5.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.5.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.6.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.6.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.7.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.7.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.8.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.8.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.9.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.experts.9.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.gate.key.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.gate.query.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.mlp.gate.value.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.self_attn.lf_gate.conv1.bias": "pytorch_model-00001-of-00009.bin", "model.layers.1.self_attn.lf_gate.conv1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.self_attn.lf_gate.conv2.bias": "pytorch_model-00001-of-00009.bin", "model.layers.1.self_attn.lf_gate.conv2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00009.bin", "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00009.bin", "model.layers.10.input_layernorm.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.0.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.0.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.1.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.1.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.10.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.10.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.11.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.11.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.12.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.12.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.13.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.13.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.14.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.14.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.15.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.15.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.16.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.16.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.17.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.17.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.18.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.18.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.19.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.19.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.2.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.2.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.20.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.20.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.21.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.21.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.22.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.22.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.23.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.23.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.24.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.24.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.25.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.25.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.26.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.26.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.27.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.27.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.28.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.28.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.29.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.29.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.3.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.3.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.30.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.30.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.31.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.31.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.4.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.4.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.5.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.5.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.6.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.6.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.7.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.7.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.8.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.8.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.9.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.experts.9.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.gate.key.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.gate.query.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.mlp.gate.value.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.self_attn.lf_gate.conv1.bias": "pytorch_model-00004-of-00009.bin", "model.layers.10.self_attn.lf_gate.conv1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.self_attn.lf_gate.conv2.bias": "pytorch_model-00004-of-00009.bin", "model.layers.10.self_attn.lf_gate.conv2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00004-of-00009.bin", "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.input_layernorm.weight": "pytorch_model-00005-of-00009.bin", "model.layers.11.mlp.experts.0.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.0.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.1.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.1.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.10.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.10.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.11.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.11.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.12.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.12.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.13.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.13.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.14.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.14.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.15.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.15.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.16.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.16.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.17.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.17.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.18.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.18.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.19.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.19.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.2.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.2.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.20.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.20.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.21.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.21.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.22.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.22.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.23.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.23.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.24.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.24.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.25.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.25.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.26.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.26.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.27.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.27.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.28.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.28.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.11.mlp.experts.29.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.11.mlp.experts.29.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.11.mlp.experts.3.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.3.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.30.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.11.mlp.experts.30.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.11.mlp.experts.31.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.11.mlp.experts.31.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.11.mlp.experts.4.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.4.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.5.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.5.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.6.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.6.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.7.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.7.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.8.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.8.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.9.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.experts.9.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.gate.key.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.gate.query.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.mlp.gate.value.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00005-of-00009.bin", "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.self_attn.lf_gate.conv1.bias": "pytorch_model-00004-of-00009.bin", "model.layers.11.self_attn.lf_gate.conv1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.self_attn.lf_gate.conv2.bias": "pytorch_model-00004-of-00009.bin", "model.layers.11.self_attn.lf_gate.conv2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00004-of-00009.bin", "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00004-of-00009.bin", "model.layers.12.input_layernorm.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.0.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.0.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.1.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.1.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.10.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.10.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.11.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.11.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.12.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.12.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.13.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.13.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.14.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.14.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.15.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.15.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.16.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.16.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.17.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.17.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.18.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.18.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.19.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.19.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.2.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.2.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.20.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.20.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.21.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.21.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.22.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.22.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.23.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.23.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.24.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.24.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.25.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.25.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.26.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.26.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.27.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.27.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.28.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.28.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.29.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.29.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.3.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.3.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.30.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.30.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.31.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.31.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.4.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.4.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.5.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.5.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.6.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.6.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.7.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.7.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.8.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.8.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.9.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.experts.9.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.gate.key.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.gate.query.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.mlp.gate.value.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.self_attn.lf_gate.conv1.bias": "pytorch_model-00005-of-00009.bin", "model.layers.12.self_attn.lf_gate.conv1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.self_attn.lf_gate.conv2.bias": "pytorch_model-00005-of-00009.bin", "model.layers.12.self_attn.lf_gate.conv2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00005-of-00009.bin", "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.input_layernorm.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.0.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.0.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.1.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.1.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.10.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.10.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.11.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.11.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.12.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.12.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.13.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.13.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.14.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.14.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.15.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.15.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.16.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.16.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.17.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.17.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.18.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.18.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.19.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.19.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.2.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.2.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.20.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.20.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.21.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.21.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.22.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.22.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.23.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.23.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.24.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.24.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.25.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.25.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.26.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.26.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.27.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.27.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.28.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.28.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.29.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.29.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.3.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.3.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.30.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.30.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.31.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.31.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.4.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.4.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.5.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.5.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.6.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.6.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.7.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.7.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.8.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.8.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.9.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.experts.9.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.gate.key.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.gate.query.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.mlp.gate.value.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.self_attn.lf_gate.conv1.bias": "pytorch_model-00005-of-00009.bin", "model.layers.13.self_attn.lf_gate.conv1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.self_attn.lf_gate.conv2.bias": "pytorch_model-00005-of-00009.bin", "model.layers.13.self_attn.lf_gate.conv2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00005-of-00009.bin", "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.input_layernorm.weight": "pytorch_model-00006-of-00009.bin", "model.layers.14.mlp.experts.0.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.0.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.1.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.1.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.10.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.10.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.11.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.11.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.12.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.12.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.13.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.13.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.14.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.14.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.15.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.15.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.16.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.16.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.17.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.17.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.18.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.18.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.19.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.19.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.2.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.2.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.20.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.20.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.21.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.21.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.22.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.22.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.23.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.23.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.24.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.24.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.25.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.25.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.26.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.26.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.27.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.27.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.28.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.28.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.29.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.14.mlp.experts.29.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.14.mlp.experts.3.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.3.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.30.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.14.mlp.experts.30.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.14.mlp.experts.31.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.14.mlp.experts.31.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.14.mlp.experts.4.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.4.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.5.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.5.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.6.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.6.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.7.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.7.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.8.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.8.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.9.w1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.experts.9.w2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.gate.key.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.gate.query.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.mlp.gate.value.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00006-of-00009.bin", "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.self_attn.lf_gate.conv1.bias": "pytorch_model-00005-of-00009.bin", "model.layers.14.self_attn.lf_gate.conv1.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.self_attn.lf_gate.conv2.bias": "pytorch_model-00005-of-00009.bin", "model.layers.14.self_attn.lf_gate.conv2.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00005-of-00009.bin", "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00005-of-00009.bin", "model.layers.15.input_layernorm.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.0.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.0.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.1.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.1.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.10.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.10.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.11.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.11.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.12.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.12.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.13.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.13.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.14.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.14.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.15.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.15.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.16.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.16.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.17.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.17.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.18.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.18.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.19.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.19.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.2.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.2.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.20.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.20.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.21.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.21.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.22.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.22.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.23.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.23.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.24.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.24.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.25.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.25.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.26.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.26.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.27.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.27.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.28.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.28.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.29.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.29.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.3.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.3.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.30.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.30.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.31.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.31.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.4.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.4.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.5.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.5.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.6.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.6.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.7.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.7.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.8.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.8.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.9.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.experts.9.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.gate.key.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.gate.query.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.mlp.gate.value.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.self_attn.lf_gate.conv1.bias": "pytorch_model-00006-of-00009.bin", "model.layers.15.self_attn.lf_gate.conv1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.self_attn.lf_gate.conv2.bias": "pytorch_model-00006-of-00009.bin", "model.layers.15.self_attn.lf_gate.conv2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00006-of-00009.bin", "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.input_layernorm.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.0.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.0.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.1.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.1.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.10.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.10.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.11.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.11.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.12.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.12.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.13.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.13.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.14.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.14.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.15.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.15.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.16.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.16.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.17.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.17.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.18.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.18.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.19.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.19.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.2.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.2.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.20.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.20.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.21.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.21.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.22.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.22.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.23.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.23.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.24.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.24.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.25.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.25.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.26.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.26.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.27.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.27.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.28.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.28.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.29.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.29.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.3.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.3.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.30.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.30.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.31.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.31.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.4.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.4.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.5.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.5.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.6.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.6.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.7.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.7.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.8.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.8.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.9.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.experts.9.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.gate.key.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.gate.query.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.mlp.gate.value.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.self_attn.lf_gate.conv1.bias": "pytorch_model-00006-of-00009.bin", "model.layers.16.self_attn.lf_gate.conv1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.self_attn.lf_gate.conv2.bias": "pytorch_model-00006-of-00009.bin", "model.layers.16.self_attn.lf_gate.conv2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00006-of-00009.bin", "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.input_layernorm.weight": "pytorch_model-00007-of-00009.bin", "model.layers.17.mlp.experts.0.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.0.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.1.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.1.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.10.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.10.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.11.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.11.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.12.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.12.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.13.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.13.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.14.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.14.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.15.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.15.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.16.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.16.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.17.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.17.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.18.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.18.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.19.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.19.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.2.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.2.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.20.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.20.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.21.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.21.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.22.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.22.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.23.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.23.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.24.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.24.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.25.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.25.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.26.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.26.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.27.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.27.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.28.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.28.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.29.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.29.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.17.mlp.experts.3.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.3.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.30.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.17.mlp.experts.30.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.17.mlp.experts.31.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.17.mlp.experts.31.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.17.mlp.experts.4.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.4.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.5.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.5.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.6.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.6.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.7.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.7.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.8.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.8.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.9.w1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.experts.9.w2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.gate.key.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.gate.query.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.mlp.gate.value.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00007-of-00009.bin", "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.self_attn.lf_gate.conv1.bias": "pytorch_model-00006-of-00009.bin", "model.layers.17.self_attn.lf_gate.conv1.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.self_attn.lf_gate.conv2.bias": "pytorch_model-00006-of-00009.bin", "model.layers.17.self_attn.lf_gate.conv2.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00006-of-00009.bin", "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00006-of-00009.bin", "model.layers.18.input_layernorm.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.0.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.0.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.1.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.1.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.10.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.10.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.11.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.11.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.12.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.12.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.13.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.13.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.14.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.14.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.15.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.15.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.16.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.16.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.17.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.17.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.18.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.18.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.19.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.19.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.2.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.2.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.20.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.20.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.21.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.21.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.22.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.22.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.23.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.23.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.24.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.24.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.25.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.25.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.26.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.26.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.27.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.27.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.28.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.28.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.29.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.29.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.3.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.3.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.30.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.30.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.31.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.31.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.4.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.4.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.5.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.5.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.6.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.6.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.7.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.7.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.8.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.8.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.9.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.experts.9.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.gate.key.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.gate.query.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.mlp.gate.value.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.self_attn.lf_gate.conv1.bias": "pytorch_model-00007-of-00009.bin", "model.layers.18.self_attn.lf_gate.conv1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.self_attn.lf_gate.conv2.bias": "pytorch_model-00007-of-00009.bin", "model.layers.18.self_attn.lf_gate.conv2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00007-of-00009.bin", "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.input_layernorm.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.0.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.0.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.1.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.1.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.10.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.10.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.11.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.11.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.12.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.12.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.13.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.13.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.14.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.14.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.15.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.15.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.16.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.16.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.17.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.17.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.18.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.18.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.19.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.19.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.2.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.2.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.20.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.20.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.21.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.21.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.22.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.22.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.23.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.23.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.24.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.24.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.25.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.25.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.26.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.26.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.27.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.27.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.28.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.28.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.29.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.29.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.3.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.3.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.30.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.30.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.31.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.31.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.4.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.4.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.5.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.5.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.6.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.6.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.7.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.7.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.8.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.8.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.9.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.experts.9.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.gate.key.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.gate.query.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.mlp.gate.value.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.self_attn.lf_gate.conv1.bias": "pytorch_model-00007-of-00009.bin", "model.layers.19.self_attn.lf_gate.conv1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.self_attn.lf_gate.conv2.bias": "pytorch_model-00007-of-00009.bin", "model.layers.19.self_attn.lf_gate.conv2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00007-of-00009.bin", "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00007-of-00009.bin", "model.layers.2.input_layernorm.weight": "pytorch_model-00002-of-00009.bin", "model.layers.2.mlp.experts.0.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.0.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.1.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.1.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.10.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.10.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.11.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.11.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.12.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.12.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.13.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.13.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.14.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.14.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.15.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.15.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.16.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.16.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.17.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.17.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.18.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.18.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.19.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.19.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.2.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.2.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.20.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.20.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.21.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.21.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.22.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.22.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.23.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.23.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.24.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.24.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.25.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.25.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.26.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.26.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.27.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.2.mlp.experts.27.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.2.mlp.experts.28.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.2.mlp.experts.28.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.2.mlp.experts.29.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.2.mlp.experts.29.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.2.mlp.experts.3.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.3.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.30.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.2.mlp.experts.30.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.2.mlp.experts.31.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.2.mlp.experts.31.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.2.mlp.experts.4.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.4.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.5.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.5.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.6.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.6.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.7.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.7.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.8.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.8.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.9.w1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.experts.9.w2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.gate.key.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.gate.query.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.mlp.gate.value.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00002-of-00009.bin", "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.self_attn.lf_gate.conv1.bias": "pytorch_model-00001-of-00009.bin", "model.layers.2.self_attn.lf_gate.conv1.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.self_attn.lf_gate.conv2.bias": "pytorch_model-00001-of-00009.bin", "model.layers.2.self_attn.lf_gate.conv2.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00009.bin", "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00009.bin", "model.layers.20.input_layernorm.weight": "pytorch_model-00008-of-00009.bin", "model.layers.20.mlp.experts.0.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.0.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.1.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.1.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.10.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.10.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.11.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.11.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.12.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.12.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.13.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.13.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.14.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.14.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.15.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.15.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.16.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.16.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.17.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.17.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.18.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.18.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.19.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.19.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.2.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.2.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.20.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.20.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.21.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.21.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.22.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.22.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.23.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.23.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.24.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.24.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.25.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.25.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.26.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.26.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.27.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.27.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.28.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.28.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.29.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.29.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.3.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.3.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.30.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.20.mlp.experts.30.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.20.mlp.experts.31.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.20.mlp.experts.31.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.20.mlp.experts.4.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.4.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.5.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.5.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.6.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.6.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.7.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.7.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.8.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.8.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.9.w1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.experts.9.w2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.gate.key.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.gate.query.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.mlp.gate.value.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00008-of-00009.bin", "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.self_attn.lf_gate.conv1.bias": "pytorch_model-00007-of-00009.bin", "model.layers.20.self_attn.lf_gate.conv1.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.self_attn.lf_gate.conv2.bias": "pytorch_model-00007-of-00009.bin", "model.layers.20.self_attn.lf_gate.conv2.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00007-of-00009.bin", "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00007-of-00009.bin", "model.layers.21.input_layernorm.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.0.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.0.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.1.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.1.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.10.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.10.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.11.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.11.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.12.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.12.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.13.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.13.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.14.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.14.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.15.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.15.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.16.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.16.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.17.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.17.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.18.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.18.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.19.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.19.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.2.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.2.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.20.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.20.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.21.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.21.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.22.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.22.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.23.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.23.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.24.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.24.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.25.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.25.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.26.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.26.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.27.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.27.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.28.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.28.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.29.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.29.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.3.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.3.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.30.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.30.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.31.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.31.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.4.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.4.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.5.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.5.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.6.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.6.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.7.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.7.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.8.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.8.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.9.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.experts.9.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.gate.key.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.gate.query.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.mlp.gate.value.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.self_attn.lf_gate.conv1.bias": "pytorch_model-00008-of-00009.bin", "model.layers.21.self_attn.lf_gate.conv1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.self_attn.lf_gate.conv2.bias": "pytorch_model-00008-of-00009.bin", "model.layers.21.self_attn.lf_gate.conv2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00008-of-00009.bin", "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.input_layernorm.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.0.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.0.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.1.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.1.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.10.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.10.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.11.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.11.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.12.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.12.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.13.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.13.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.14.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.14.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.15.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.15.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.16.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.16.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.17.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.17.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.18.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.18.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.19.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.19.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.2.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.2.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.20.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.20.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.21.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.21.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.22.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.22.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.23.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.23.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.24.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.24.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.25.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.25.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.26.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.26.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.27.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.27.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.28.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.28.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.29.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.29.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.3.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.3.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.30.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.30.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.31.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.31.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.4.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.4.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.5.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.5.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.6.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.6.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.7.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.7.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.8.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.8.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.9.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.experts.9.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.gate.key.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.gate.query.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.mlp.gate.value.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.self_attn.lf_gate.conv1.bias": "pytorch_model-00008-of-00009.bin", "model.layers.22.self_attn.lf_gate.conv1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.self_attn.lf_gate.conv2.bias": "pytorch_model-00008-of-00009.bin", "model.layers.22.self_attn.lf_gate.conv2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00008-of-00009.bin", "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.input_layernorm.weight": "pytorch_model-00009-of-00009.bin", "model.layers.23.mlp.experts.0.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.0.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.1.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.1.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.10.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.10.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.11.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.11.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.12.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.12.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.13.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.13.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.14.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.14.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.15.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.15.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.16.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.16.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.17.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.17.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.18.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.18.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.19.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.19.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.2.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.2.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.20.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.20.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.21.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.21.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.22.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.22.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.23.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.23.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.24.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.24.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.25.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.25.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.26.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.26.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.27.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.27.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.28.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.28.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.29.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.29.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.3.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.3.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.30.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.30.w2.weight": "pytorch_model-00009-of-00009.bin", "model.layers.23.mlp.experts.31.w1.weight": "pytorch_model-00009-of-00009.bin", "model.layers.23.mlp.experts.31.w2.weight": "pytorch_model-00009-of-00009.bin", "model.layers.23.mlp.experts.4.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.4.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.5.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.5.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.6.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.6.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.7.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.7.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.8.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.8.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.9.w1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.experts.9.w2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.gate.key.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.gate.query.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.mlp.gate.value.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00009-of-00009.bin", "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.self_attn.lf_gate.conv1.bias": "pytorch_model-00008-of-00009.bin", "model.layers.23.self_attn.lf_gate.conv1.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.self_attn.lf_gate.conv2.bias": "pytorch_model-00008-of-00009.bin", "model.layers.23.self_attn.lf_gate.conv2.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00008-of-00009.bin", "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00008-of-00009.bin", "model.layers.3.input_layernorm.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.0.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.0.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.1.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.1.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.10.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.10.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.11.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.11.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.12.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.12.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.13.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.13.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.14.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.14.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.15.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.15.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.16.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.16.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.17.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.17.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.18.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.18.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.19.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.19.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.2.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.2.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.20.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.20.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.21.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.21.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.22.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.22.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.23.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.23.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.24.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.24.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.25.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.25.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.26.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.26.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.27.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.27.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.28.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.28.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.29.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.29.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.3.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.3.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.30.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.30.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.31.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.31.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.4.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.4.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.5.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.5.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.6.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.6.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.7.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.7.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.8.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.8.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.9.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.experts.9.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.gate.key.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.gate.query.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.mlp.gate.value.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.self_attn.lf_gate.conv1.bias": "pytorch_model-00002-of-00009.bin", "model.layers.3.self_attn.lf_gate.conv1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.self_attn.lf_gate.conv2.bias": "pytorch_model-00002-of-00009.bin", "model.layers.3.self_attn.lf_gate.conv2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00002-of-00009.bin", "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.input_layernorm.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.0.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.0.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.1.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.1.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.10.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.10.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.11.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.11.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.12.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.12.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.13.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.13.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.14.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.14.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.15.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.15.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.16.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.16.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.17.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.17.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.18.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.18.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.19.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.19.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.2.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.2.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.20.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.20.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.21.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.21.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.22.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.22.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.23.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.23.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.24.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.24.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.25.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.25.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.26.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.26.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.27.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.27.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.28.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.28.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.29.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.29.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.3.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.3.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.30.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.30.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.31.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.31.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.4.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.4.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.5.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.5.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.6.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.6.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.7.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.7.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.8.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.8.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.9.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.experts.9.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.gate.key.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.gate.query.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.mlp.gate.value.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.self_attn.lf_gate.conv1.bias": "pytorch_model-00002-of-00009.bin", "model.layers.4.self_attn.lf_gate.conv1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.self_attn.lf_gate.conv2.bias": "pytorch_model-00002-of-00009.bin", "model.layers.4.self_attn.lf_gate.conv2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00002-of-00009.bin", "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.input_layernorm.weight": "pytorch_model-00003-of-00009.bin", "model.layers.5.mlp.experts.0.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.0.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.1.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.1.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.10.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.10.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.11.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.11.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.12.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.12.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.13.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.13.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.14.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.14.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.15.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.15.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.16.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.16.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.17.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.17.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.18.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.18.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.19.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.19.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.2.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.2.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.20.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.20.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.21.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.21.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.22.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.22.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.23.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.23.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.24.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.24.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.25.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.25.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.26.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.26.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.27.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.27.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.5.mlp.experts.28.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.5.mlp.experts.28.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.5.mlp.experts.29.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.5.mlp.experts.29.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.5.mlp.experts.3.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.3.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.30.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.5.mlp.experts.30.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.5.mlp.experts.31.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.5.mlp.experts.31.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.5.mlp.experts.4.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.4.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.5.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.5.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.6.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.6.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.7.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.7.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.8.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.8.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.9.w1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.experts.9.w2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.gate.key.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.gate.query.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.mlp.gate.value.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00003-of-00009.bin", "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.self_attn.lf_gate.conv1.bias": "pytorch_model-00002-of-00009.bin", "model.layers.5.self_attn.lf_gate.conv1.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.self_attn.lf_gate.conv2.bias": "pytorch_model-00002-of-00009.bin", "model.layers.5.self_attn.lf_gate.conv2.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00002-of-00009.bin", "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00002-of-00009.bin", "model.layers.6.input_layernorm.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.0.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.0.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.1.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.1.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.10.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.10.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.11.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.11.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.12.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.12.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.13.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.13.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.14.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.14.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.15.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.15.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.16.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.16.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.17.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.17.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.18.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.18.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.19.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.19.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.2.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.2.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.20.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.20.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.21.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.21.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.22.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.22.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.23.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.23.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.24.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.24.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.25.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.25.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.26.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.26.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.27.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.27.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.28.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.28.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.29.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.29.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.3.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.3.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.30.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.30.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.31.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.31.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.4.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.4.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.5.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.5.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.6.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.6.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.7.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.7.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.8.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.8.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.9.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.experts.9.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.gate.key.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.gate.query.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.mlp.gate.value.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.self_attn.lf_gate.conv1.bias": "pytorch_model-00003-of-00009.bin", "model.layers.6.self_attn.lf_gate.conv1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.self_attn.lf_gate.conv2.bias": "pytorch_model-00003-of-00009.bin", "model.layers.6.self_attn.lf_gate.conv2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00003-of-00009.bin", "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.input_layernorm.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.0.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.0.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.1.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.1.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.10.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.10.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.11.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.11.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.12.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.12.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.13.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.13.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.14.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.14.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.15.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.15.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.16.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.16.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.17.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.17.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.18.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.18.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.19.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.19.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.2.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.2.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.20.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.20.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.21.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.21.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.22.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.22.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.23.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.23.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.24.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.24.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.25.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.25.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.26.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.26.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.27.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.27.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.28.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.28.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.29.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.29.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.3.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.3.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.30.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.30.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.31.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.31.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.4.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.4.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.5.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.5.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.6.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.6.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.7.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.7.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.8.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.8.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.9.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.experts.9.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.gate.key.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.gate.query.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.mlp.gate.value.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.self_attn.lf_gate.conv1.bias": "pytorch_model-00003-of-00009.bin", "model.layers.7.self_attn.lf_gate.conv1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.self_attn.lf_gate.conv2.bias": "pytorch_model-00003-of-00009.bin", "model.layers.7.self_attn.lf_gate.conv2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00003-of-00009.bin", "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.input_layernorm.weight": "pytorch_model-00004-of-00009.bin", "model.layers.8.mlp.experts.0.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.0.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.1.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.1.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.10.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.10.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.11.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.11.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.12.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.12.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.13.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.13.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.14.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.14.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.15.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.15.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.16.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.16.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.17.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.17.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.18.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.18.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.19.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.19.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.2.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.2.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.20.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.20.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.21.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.21.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.22.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.22.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.23.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.23.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.24.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.24.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.25.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.25.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.26.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.26.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.27.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.27.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.28.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.8.mlp.experts.28.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.8.mlp.experts.29.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.8.mlp.experts.29.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.8.mlp.experts.3.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.3.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.30.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.8.mlp.experts.30.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.8.mlp.experts.31.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.8.mlp.experts.31.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.8.mlp.experts.4.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.4.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.5.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.5.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.6.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.6.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.7.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.7.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.8.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.8.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.9.w1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.experts.9.w2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.gate.key.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.gate.query.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.mlp.gate.value.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00004-of-00009.bin", "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.self_attn.lf_gate.conv1.bias": "pytorch_model-00003-of-00009.bin", "model.layers.8.self_attn.lf_gate.conv1.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.self_attn.lf_gate.conv2.bias": "pytorch_model-00003-of-00009.bin", "model.layers.8.self_attn.lf_gate.conv2.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00003-of-00009.bin", "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00003-of-00009.bin", "model.layers.9.input_layernorm.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.0.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.0.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.1.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.1.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.10.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.10.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.11.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.11.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.12.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.12.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.13.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.13.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.14.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.14.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.15.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.15.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.16.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.16.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.17.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.17.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.18.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.18.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.19.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.19.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.2.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.2.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.20.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.20.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.21.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.21.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.22.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.22.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.23.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.23.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.24.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.24.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.25.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.25.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.26.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.26.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.27.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.27.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.28.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.28.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.29.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.29.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.3.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.3.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.30.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.30.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.31.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.31.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.4.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.4.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.5.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.5.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.6.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.6.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.7.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.7.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.8.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.8.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.9.w1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.experts.9.w2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.gate.key.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.gate.query.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.mlp.gate.value.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.self_attn.lf_gate.conv1.bias": "pytorch_model-00004-of-00009.bin", "model.layers.9.self_attn.lf_gate.conv1.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.self_attn.lf_gate.conv2.bias": "pytorch_model-00004-of-00009.bin", "model.layers.9.self_attn.lf_gate.conv2.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.self_attn.lf_gate.output_layernorm.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00004-of-00009.bin", "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00004-of-00009.bin", "model.norm.weight": "pytorch_model-00009-of-00009.bin" } }