{ "metadata": { "total_size": 95374645760 }, "weight_map": { "lm_head.weight": "pytorch_model-00001-of-00020.bin", "model.embed_tokens.weight": "pytorch_model-00001-of-00020.bin", "model.embedding_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.0.conv.conv.weight": "pytorch_model-00001-of-00020.bin", "model.layers.0.conv.in_proj.weight": "pytorch_model-00001-of-00020.bin", "model.layers.0.conv.out_proj.weight": "pytorch_model-00001-of-00020.bin", "model.layers.0.feed_forward.w1.weight": "pytorch_model-00001-of-00020.bin", "model.layers.0.feed_forward.w2.weight": "pytorch_model-00001-of-00020.bin", "model.layers.0.feed_forward.w3.weight": "pytorch_model-00001-of-00020.bin", "model.layers.0.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.0.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.1.conv.conv.weight": "pytorch_model-00001-of-00020.bin", "model.layers.1.conv.in_proj.weight": "pytorch_model-00001-of-00020.bin", "model.layers.1.conv.out_proj.weight": "pytorch_model-00001-of-00020.bin", "model.layers.1.feed_forward.w1.weight": "pytorch_model-00001-of-00020.bin", "model.layers.1.feed_forward.w2.weight": "pytorch_model-00001-of-00020.bin", "model.layers.1.feed_forward.w3.weight": "pytorch_model-00001-of-00020.bin", "model.layers.1.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.1.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.10.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.10.feed_forward.experts.down_proj": "pytorch_model-00005-of-00020.bin", "model.layers.10.feed_forward.experts.gate_up_proj": "pytorch_model-00005-of-00020.bin", "model.layers.10.feed_forward.gate.weight": "pytorch_model-00005-of-00020.bin", "model.layers.10.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.10.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.10.self_attn.k_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00005-of-00020.bin", "model.layers.10.self_attn.out_proj.weight": "pytorch_model-00005-of-00020.bin", "model.layers.10.self_attn.q_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00005-of-00020.bin", "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00005-of-00020.bin", "model.layers.11.conv.conv.weight": "pytorch_model-00005-of-00020.bin", "model.layers.11.conv.in_proj.weight": "pytorch_model-00005-of-00020.bin", "model.layers.11.conv.out_proj.weight": "pytorch_model-00005-of-00020.bin", "model.layers.11.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.11.feed_forward.experts.down_proj": "pytorch_model-00006-of-00020.bin", "model.layers.11.feed_forward.experts.gate_up_proj": "pytorch_model-00006-of-00020.bin", "model.layers.11.feed_forward.gate.weight": "pytorch_model-00005-of-00020.bin", "model.layers.11.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.11.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.12.conv.conv.weight": "pytorch_model-00006-of-00020.bin", "model.layers.12.conv.in_proj.weight": "pytorch_model-00006-of-00020.bin", "model.layers.12.conv.out_proj.weight": "pytorch_model-00006-of-00020.bin", "model.layers.12.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.12.feed_forward.experts.down_proj": "pytorch_model-00006-of-00020.bin", "model.layers.12.feed_forward.experts.gate_up_proj": "pytorch_model-00006-of-00020.bin", "model.layers.12.feed_forward.gate.weight": "pytorch_model-00006-of-00020.bin", "model.layers.12.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.12.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.13.conv.conv.weight": "pytorch_model-00006-of-00020.bin", "model.layers.13.conv.in_proj.weight": "pytorch_model-00006-of-00020.bin", "model.layers.13.conv.out_proj.weight": "pytorch_model-00006-of-00020.bin", "model.layers.13.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.13.feed_forward.experts.down_proj": "pytorch_model-00007-of-00020.bin", "model.layers.13.feed_forward.experts.gate_up_proj": "pytorch_model-00007-of-00020.bin", "model.layers.13.feed_forward.gate.weight": "pytorch_model-00006-of-00020.bin", "model.layers.13.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.13.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.14.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.14.feed_forward.experts.down_proj": "pytorch_model-00007-of-00020.bin", "model.layers.14.feed_forward.experts.gate_up_proj": "pytorch_model-00007-of-00020.bin", "model.layers.14.feed_forward.gate.weight": "pytorch_model-00007-of-00020.bin", "model.layers.14.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.14.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.14.self_attn.k_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00007-of-00020.bin", "model.layers.14.self_attn.out_proj.weight": "pytorch_model-00007-of-00020.bin", "model.layers.14.self_attn.q_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00007-of-00020.bin", "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00007-of-00020.bin", "model.layers.15.conv.conv.weight": "pytorch_model-00007-of-00020.bin", "model.layers.15.conv.in_proj.weight": "pytorch_model-00007-of-00020.bin", "model.layers.15.conv.out_proj.weight": "pytorch_model-00007-of-00020.bin", "model.layers.15.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.15.feed_forward.experts.down_proj": "pytorch_model-00008-of-00020.bin", "model.layers.15.feed_forward.experts.gate_up_proj": "pytorch_model-00008-of-00020.bin", "model.layers.15.feed_forward.gate.weight": "pytorch_model-00007-of-00020.bin", "model.layers.15.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.15.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.16.conv.conv.weight": "pytorch_model-00008-of-00020.bin", "model.layers.16.conv.in_proj.weight": "pytorch_model-00008-of-00020.bin", "model.layers.16.conv.out_proj.weight": "pytorch_model-00008-of-00020.bin", "model.layers.16.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.16.feed_forward.experts.down_proj": "pytorch_model-00008-of-00020.bin", "model.layers.16.feed_forward.experts.gate_up_proj": "pytorch_model-00008-of-00020.bin", "model.layers.16.feed_forward.gate.weight": "pytorch_model-00008-of-00020.bin", "model.layers.16.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.16.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.17.conv.conv.weight": "pytorch_model-00008-of-00020.bin", "model.layers.17.conv.in_proj.weight": "pytorch_model-00008-of-00020.bin", "model.layers.17.conv.out_proj.weight": "pytorch_model-00008-of-00020.bin", "model.layers.17.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.17.feed_forward.experts.down_proj": "pytorch_model-00009-of-00020.bin", "model.layers.17.feed_forward.experts.gate_up_proj": "pytorch_model-00009-of-00020.bin", "model.layers.17.feed_forward.gate.weight": "pytorch_model-00008-of-00020.bin", "model.layers.17.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.17.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.18.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.18.feed_forward.experts.down_proj": "pytorch_model-00009-of-00020.bin", "model.layers.18.feed_forward.experts.gate_up_proj": "pytorch_model-00009-of-00020.bin", "model.layers.18.feed_forward.gate.weight": "pytorch_model-00009-of-00020.bin", "model.layers.18.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.18.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.18.self_attn.k_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00009-of-00020.bin", "model.layers.18.self_attn.out_proj.weight": "pytorch_model-00009-of-00020.bin", "model.layers.18.self_attn.q_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00009-of-00020.bin", "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00009-of-00020.bin", "model.layers.19.conv.conv.weight": "pytorch_model-00009-of-00020.bin", "model.layers.19.conv.in_proj.weight": "pytorch_model-00009-of-00020.bin", "model.layers.19.conv.out_proj.weight": "pytorch_model-00009-of-00020.bin", "model.layers.19.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.19.feed_forward.experts.down_proj": "pytorch_model-00010-of-00020.bin", "model.layers.19.feed_forward.experts.gate_up_proj": "pytorch_model-00010-of-00020.bin", "model.layers.19.feed_forward.gate.weight": "pytorch_model-00009-of-00020.bin", "model.layers.19.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.19.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.2.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.2.feed_forward.experts.down_proj": "pytorch_model-00001-of-00020.bin", "model.layers.2.feed_forward.experts.gate_up_proj": "pytorch_model-00001-of-00020.bin", "model.layers.2.feed_forward.gate.weight": "pytorch_model-00001-of-00020.bin", "model.layers.2.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.2.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.2.self_attn.k_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00020.bin", "model.layers.2.self_attn.out_proj.weight": "pytorch_model-00001-of-00020.bin", "model.layers.2.self_attn.q_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00020.bin", "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00020.bin", "model.layers.20.conv.conv.weight": "pytorch_model-00010-of-00020.bin", "model.layers.20.conv.in_proj.weight": "pytorch_model-00010-of-00020.bin", "model.layers.20.conv.out_proj.weight": "pytorch_model-00010-of-00020.bin", "model.layers.20.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.20.feed_forward.experts.down_proj": "pytorch_model-00010-of-00020.bin", "model.layers.20.feed_forward.experts.gate_up_proj": "pytorch_model-00010-of-00020.bin", "model.layers.20.feed_forward.gate.weight": "pytorch_model-00010-of-00020.bin", "model.layers.20.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.20.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.21.conv.conv.weight": "pytorch_model-00010-of-00020.bin", "model.layers.21.conv.in_proj.weight": "pytorch_model-00010-of-00020.bin", "model.layers.21.conv.out_proj.weight": "pytorch_model-00010-of-00020.bin", "model.layers.21.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.21.feed_forward.experts.down_proj": "pytorch_model-00011-of-00020.bin", "model.layers.21.feed_forward.experts.gate_up_proj": "pytorch_model-00011-of-00020.bin", "model.layers.21.feed_forward.gate.weight": "pytorch_model-00010-of-00020.bin", "model.layers.21.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.21.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.22.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.22.feed_forward.experts.down_proj": "pytorch_model-00011-of-00020.bin", "model.layers.22.feed_forward.experts.gate_up_proj": "pytorch_model-00011-of-00020.bin", "model.layers.22.feed_forward.gate.weight": "pytorch_model-00011-of-00020.bin", "model.layers.22.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.22.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.22.self_attn.k_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00011-of-00020.bin", "model.layers.22.self_attn.out_proj.weight": "pytorch_model-00011-of-00020.bin", "model.layers.22.self_attn.q_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00011-of-00020.bin", "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00011-of-00020.bin", "model.layers.23.conv.conv.weight": "pytorch_model-00011-of-00020.bin", "model.layers.23.conv.in_proj.weight": "pytorch_model-00011-of-00020.bin", "model.layers.23.conv.out_proj.weight": "pytorch_model-00011-of-00020.bin", "model.layers.23.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.23.feed_forward.experts.down_proj": "pytorch_model-00012-of-00020.bin", "model.layers.23.feed_forward.experts.gate_up_proj": "pytorch_model-00012-of-00020.bin", "model.layers.23.feed_forward.gate.weight": "pytorch_model-00011-of-00020.bin", "model.layers.23.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.23.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.24.conv.conv.weight": "pytorch_model-00012-of-00020.bin", "model.layers.24.conv.in_proj.weight": "pytorch_model-00012-of-00020.bin", "model.layers.24.conv.out_proj.weight": "pytorch_model-00012-of-00020.bin", "model.layers.24.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.24.feed_forward.experts.down_proj": "pytorch_model-00012-of-00020.bin", "model.layers.24.feed_forward.experts.gate_up_proj": "pytorch_model-00012-of-00020.bin", "model.layers.24.feed_forward.gate.weight": "pytorch_model-00012-of-00020.bin", "model.layers.24.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.24.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.25.conv.conv.weight": "pytorch_model-00012-of-00020.bin", "model.layers.25.conv.in_proj.weight": "pytorch_model-00012-of-00020.bin", "model.layers.25.conv.out_proj.weight": "pytorch_model-00012-of-00020.bin", "model.layers.25.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.25.feed_forward.experts.down_proj": "pytorch_model-00013-of-00020.bin", "model.layers.25.feed_forward.experts.gate_up_proj": "pytorch_model-00013-of-00020.bin", "model.layers.25.feed_forward.gate.weight": "pytorch_model-00012-of-00020.bin", "model.layers.25.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.25.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.26.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.26.feed_forward.experts.down_proj": "pytorch_model-00013-of-00020.bin", "model.layers.26.feed_forward.experts.gate_up_proj": "pytorch_model-00013-of-00020.bin", "model.layers.26.feed_forward.gate.weight": "pytorch_model-00013-of-00020.bin", "model.layers.26.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.26.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.26.self_attn.k_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00013-of-00020.bin", "model.layers.26.self_attn.out_proj.weight": "pytorch_model-00013-of-00020.bin", "model.layers.26.self_attn.q_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00013-of-00020.bin", "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00013-of-00020.bin", "model.layers.27.conv.conv.weight": "pytorch_model-00013-of-00020.bin", "model.layers.27.conv.in_proj.weight": "pytorch_model-00013-of-00020.bin", "model.layers.27.conv.out_proj.weight": "pytorch_model-00013-of-00020.bin", "model.layers.27.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.27.feed_forward.experts.down_proj": "pytorch_model-00014-of-00020.bin", "model.layers.27.feed_forward.experts.gate_up_proj": "pytorch_model-00014-of-00020.bin", "model.layers.27.feed_forward.gate.weight": "pytorch_model-00013-of-00020.bin", "model.layers.27.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.27.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.28.conv.conv.weight": "pytorch_model-00014-of-00020.bin", "model.layers.28.conv.in_proj.weight": "pytorch_model-00014-of-00020.bin", "model.layers.28.conv.out_proj.weight": "pytorch_model-00014-of-00020.bin", "model.layers.28.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.28.feed_forward.experts.down_proj": "pytorch_model-00014-of-00020.bin", "model.layers.28.feed_forward.experts.gate_up_proj": "pytorch_model-00014-of-00020.bin", "model.layers.28.feed_forward.gate.weight": "pytorch_model-00014-of-00020.bin", "model.layers.28.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.28.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.29.conv.conv.weight": "pytorch_model-00014-of-00020.bin", "model.layers.29.conv.in_proj.weight": "pytorch_model-00014-of-00020.bin", "model.layers.29.conv.out_proj.weight": "pytorch_model-00014-of-00020.bin", "model.layers.29.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.29.feed_forward.experts.down_proj": "pytorch_model-00015-of-00020.bin", "model.layers.29.feed_forward.experts.gate_up_proj": "pytorch_model-00015-of-00020.bin", "model.layers.29.feed_forward.gate.weight": "pytorch_model-00014-of-00020.bin", "model.layers.29.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.29.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.3.conv.conv.weight": "pytorch_model-00001-of-00020.bin", "model.layers.3.conv.in_proj.weight": "pytorch_model-00001-of-00020.bin", "model.layers.3.conv.out_proj.weight": "pytorch_model-00001-of-00020.bin", "model.layers.3.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.3.feed_forward.experts.down_proj": "pytorch_model-00002-of-00020.bin", "model.layers.3.feed_forward.experts.gate_up_proj": "pytorch_model-00002-of-00020.bin", "model.layers.3.feed_forward.gate.weight": "pytorch_model-00001-of-00020.bin", "model.layers.3.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.3.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.30.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.30.feed_forward.experts.down_proj": "pytorch_model-00015-of-00020.bin", "model.layers.30.feed_forward.experts.gate_up_proj": "pytorch_model-00015-of-00020.bin", "model.layers.30.feed_forward.gate.weight": "pytorch_model-00015-of-00020.bin", "model.layers.30.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.30.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.30.self_attn.k_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00015-of-00020.bin", "model.layers.30.self_attn.out_proj.weight": "pytorch_model-00015-of-00020.bin", "model.layers.30.self_attn.q_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00015-of-00020.bin", "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00015-of-00020.bin", "model.layers.31.conv.conv.weight": "pytorch_model-00015-of-00020.bin", "model.layers.31.conv.in_proj.weight": "pytorch_model-00015-of-00020.bin", "model.layers.31.conv.out_proj.weight": "pytorch_model-00015-of-00020.bin", "model.layers.31.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.31.feed_forward.experts.down_proj": "pytorch_model-00016-of-00020.bin", "model.layers.31.feed_forward.experts.gate_up_proj": "pytorch_model-00016-of-00020.bin", "model.layers.31.feed_forward.gate.weight": "pytorch_model-00015-of-00020.bin", "model.layers.31.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.31.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.32.conv.conv.weight": "pytorch_model-00016-of-00020.bin", "model.layers.32.conv.in_proj.weight": "pytorch_model-00016-of-00020.bin", "model.layers.32.conv.out_proj.weight": "pytorch_model-00016-of-00020.bin", "model.layers.32.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.32.feed_forward.experts.down_proj": "pytorch_model-00016-of-00020.bin", "model.layers.32.feed_forward.experts.gate_up_proj": "pytorch_model-00016-of-00020.bin", "model.layers.32.feed_forward.gate.weight": "pytorch_model-00016-of-00020.bin", "model.layers.32.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.32.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.33.conv.conv.weight": "pytorch_model-00016-of-00020.bin", "model.layers.33.conv.in_proj.weight": "pytorch_model-00016-of-00020.bin", "model.layers.33.conv.out_proj.weight": "pytorch_model-00016-of-00020.bin", "model.layers.33.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.33.feed_forward.experts.down_proj": "pytorch_model-00017-of-00020.bin", "model.layers.33.feed_forward.experts.gate_up_proj": "pytorch_model-00017-of-00020.bin", "model.layers.33.feed_forward.gate.weight": "pytorch_model-00016-of-00020.bin", "model.layers.33.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.33.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.34.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.34.feed_forward.experts.down_proj": "pytorch_model-00017-of-00020.bin", "model.layers.34.feed_forward.experts.gate_up_proj": "pytorch_model-00017-of-00020.bin", "model.layers.34.feed_forward.gate.weight": "pytorch_model-00017-of-00020.bin", "model.layers.34.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.34.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.34.self_attn.k_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.34.self_attn.k_proj.weight": "pytorch_model-00017-of-00020.bin", "model.layers.34.self_attn.out_proj.weight": "pytorch_model-00017-of-00020.bin", "model.layers.34.self_attn.q_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.34.self_attn.q_proj.weight": "pytorch_model-00017-of-00020.bin", "model.layers.34.self_attn.v_proj.weight": "pytorch_model-00017-of-00020.bin", "model.layers.35.conv.conv.weight": "pytorch_model-00017-of-00020.bin", "model.layers.35.conv.in_proj.weight": "pytorch_model-00017-of-00020.bin", "model.layers.35.conv.out_proj.weight": "pytorch_model-00017-of-00020.bin", "model.layers.35.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.35.feed_forward.experts.down_proj": "pytorch_model-00018-of-00020.bin", "model.layers.35.feed_forward.experts.gate_up_proj": "pytorch_model-00018-of-00020.bin", "model.layers.35.feed_forward.gate.weight": "pytorch_model-00017-of-00020.bin", "model.layers.35.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.35.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.36.conv.conv.weight": "pytorch_model-00018-of-00020.bin", "model.layers.36.conv.in_proj.weight": "pytorch_model-00018-of-00020.bin", "model.layers.36.conv.out_proj.weight": "pytorch_model-00018-of-00020.bin", "model.layers.36.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.36.feed_forward.experts.down_proj": "pytorch_model-00018-of-00020.bin", "model.layers.36.feed_forward.experts.gate_up_proj": "pytorch_model-00018-of-00020.bin", "model.layers.36.feed_forward.gate.weight": "pytorch_model-00018-of-00020.bin", "model.layers.36.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.36.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.37.conv.conv.weight": "pytorch_model-00018-of-00020.bin", "model.layers.37.conv.in_proj.weight": "pytorch_model-00018-of-00020.bin", "model.layers.37.conv.out_proj.weight": "pytorch_model-00018-of-00020.bin", "model.layers.37.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.37.feed_forward.experts.down_proj": "pytorch_model-00019-of-00020.bin", "model.layers.37.feed_forward.experts.gate_up_proj": "pytorch_model-00019-of-00020.bin", "model.layers.37.feed_forward.gate.weight": "pytorch_model-00018-of-00020.bin", "model.layers.37.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.37.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.38.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.38.feed_forward.experts.down_proj": "pytorch_model-00019-of-00020.bin", "model.layers.38.feed_forward.experts.gate_up_proj": "pytorch_model-00019-of-00020.bin", "model.layers.38.feed_forward.gate.weight": "pytorch_model-00019-of-00020.bin", "model.layers.38.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.38.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.38.self_attn.k_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.38.self_attn.k_proj.weight": "pytorch_model-00019-of-00020.bin", "model.layers.38.self_attn.out_proj.weight": "pytorch_model-00019-of-00020.bin", "model.layers.38.self_attn.q_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.38.self_attn.q_proj.weight": "pytorch_model-00019-of-00020.bin", "model.layers.38.self_attn.v_proj.weight": "pytorch_model-00019-of-00020.bin", "model.layers.39.conv.conv.weight": "pytorch_model-00019-of-00020.bin", "model.layers.39.conv.in_proj.weight": "pytorch_model-00019-of-00020.bin", "model.layers.39.conv.out_proj.weight": "pytorch_model-00019-of-00020.bin", "model.layers.39.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.39.feed_forward.experts.down_proj": "pytorch_model-00020-of-00020.bin", "model.layers.39.feed_forward.experts.gate_up_proj": "pytorch_model-00020-of-00020.bin", "model.layers.39.feed_forward.gate.weight": "pytorch_model-00019-of-00020.bin", "model.layers.39.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.39.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.4.conv.conv.weight": "pytorch_model-00002-of-00020.bin", "model.layers.4.conv.in_proj.weight": "pytorch_model-00002-of-00020.bin", "model.layers.4.conv.out_proj.weight": "pytorch_model-00002-of-00020.bin", "model.layers.4.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.4.feed_forward.experts.down_proj": "pytorch_model-00002-of-00020.bin", "model.layers.4.feed_forward.experts.gate_up_proj": "pytorch_model-00002-of-00020.bin", "model.layers.4.feed_forward.gate.weight": "pytorch_model-00002-of-00020.bin", "model.layers.4.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.4.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.5.conv.conv.weight": "pytorch_model-00002-of-00020.bin", "model.layers.5.conv.in_proj.weight": "pytorch_model-00002-of-00020.bin", "model.layers.5.conv.out_proj.weight": "pytorch_model-00002-of-00020.bin", "model.layers.5.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.5.feed_forward.experts.down_proj": "pytorch_model-00003-of-00020.bin", "model.layers.5.feed_forward.experts.gate_up_proj": "pytorch_model-00003-of-00020.bin", "model.layers.5.feed_forward.gate.weight": "pytorch_model-00002-of-00020.bin", "model.layers.5.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.5.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.6.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.6.feed_forward.experts.down_proj": "pytorch_model-00003-of-00020.bin", "model.layers.6.feed_forward.experts.gate_up_proj": "pytorch_model-00003-of-00020.bin", "model.layers.6.feed_forward.gate.weight": "pytorch_model-00003-of-00020.bin", "model.layers.6.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.6.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.6.self_attn.k_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00003-of-00020.bin", "model.layers.6.self_attn.out_proj.weight": "pytorch_model-00003-of-00020.bin", "model.layers.6.self_attn.q_layernorm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00003-of-00020.bin", "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00003-of-00020.bin", "model.layers.7.conv.conv.weight": "pytorch_model-00003-of-00020.bin", "model.layers.7.conv.in_proj.weight": "pytorch_model-00003-of-00020.bin", "model.layers.7.conv.out_proj.weight": "pytorch_model-00003-of-00020.bin", "model.layers.7.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.7.feed_forward.experts.down_proj": "pytorch_model-00004-of-00020.bin", "model.layers.7.feed_forward.experts.gate_up_proj": "pytorch_model-00004-of-00020.bin", "model.layers.7.feed_forward.gate.weight": "pytorch_model-00003-of-00020.bin", "model.layers.7.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.7.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.8.conv.conv.weight": "pytorch_model-00004-of-00020.bin", "model.layers.8.conv.in_proj.weight": "pytorch_model-00004-of-00020.bin", "model.layers.8.conv.out_proj.weight": "pytorch_model-00004-of-00020.bin", "model.layers.8.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.8.feed_forward.experts.down_proj": "pytorch_model-00004-of-00020.bin", "model.layers.8.feed_forward.experts.gate_up_proj": "pytorch_model-00004-of-00020.bin", "model.layers.8.feed_forward.gate.weight": "pytorch_model-00004-of-00020.bin", "model.layers.8.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.8.operator_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.9.conv.conv.weight": "pytorch_model-00004-of-00020.bin", "model.layers.9.conv.in_proj.weight": "pytorch_model-00004-of-00020.bin", "model.layers.9.conv.out_proj.weight": "pytorch_model-00004-of-00020.bin", "model.layers.9.feed_forward.expert_bias": "pytorch_model-00001-of-00020.bin", "model.layers.9.feed_forward.experts.down_proj": "pytorch_model-00005-of-00020.bin", "model.layers.9.feed_forward.experts.gate_up_proj": "pytorch_model-00005-of-00020.bin", "model.layers.9.feed_forward.gate.weight": "pytorch_model-00004-of-00020.bin", "model.layers.9.ffn_norm.weight": "pytorch_model-00020-of-00020.bin", "model.layers.9.operator_norm.weight": "pytorch_model-00020-of-00020.bin" } }