| { |
| "metadata": { |
| "total_parameters": 3678273792, |
| "total_size": 7356547584 |
| }, |
| "weight_map": { |
| "lm_head.weight": "model-00002-of-00002.safetensors", |
| "model.embed_tokens.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.input_layernorms.0.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.input_layernorms.1.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.input_layernorms.2.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.input_layernorms.3.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.mlp.experts.down_proj": "model-00001-of-00002.safetensors", |
| "model.layers.0.mlp.experts.gate_proj": "model-00001-of-00002.safetensors", |
| "model.layers.0.mlp.experts.up_proj": "model-00001-of-00002.safetensors", |
| "model.layers.0.mlp.gate.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.post_attention_layernorms.0.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.post_attention_layernorms.1.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.post_attention_layernorms.2.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.post_attention_layernorms.3.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.0.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.0.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.0.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.0.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.0.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.0.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.1.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.1.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.1.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.1.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.1.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.1.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.2.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.2.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.2.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.2.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.2.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.2.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.3.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.3.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.3.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.3.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.3.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.0.self_attns.3.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.input_layernorms.0.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.input_layernorms.1.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.input_layernorms.2.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.input_layernorms.3.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.mlp.experts.down_proj": "model-00001-of-00002.safetensors", |
| "model.layers.1.mlp.experts.gate_proj": "model-00001-of-00002.safetensors", |
| "model.layers.1.mlp.experts.up_proj": "model-00001-of-00002.safetensors", |
| "model.layers.1.mlp.gate.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.post_attention_layernorms.0.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.post_attention_layernorms.1.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.post_attention_layernorms.2.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.post_attention_layernorms.3.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.0.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.0.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.0.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.0.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.0.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.0.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.1.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.1.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.1.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.1.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.1.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.1.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.2.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.2.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.2.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.2.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.2.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.2.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.3.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.3.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.3.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.3.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.3.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.1.self_attns.3.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.input_layernorms.0.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.input_layernorms.1.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.input_layernorms.2.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.input_layernorms.3.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.mlp.experts.down_proj": "model-00001-of-00002.safetensors", |
| "model.layers.2.mlp.experts.gate_proj": "model-00001-of-00002.safetensors", |
| "model.layers.2.mlp.experts.up_proj": "model-00001-of-00002.safetensors", |
| "model.layers.2.mlp.gate.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.post_attention_layernorms.0.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.post_attention_layernorms.1.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.post_attention_layernorms.2.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.post_attention_layernorms.3.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.0.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.0.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.0.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.0.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.0.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.0.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.1.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.1.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.1.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.1.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.1.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.1.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.2.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.2.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.2.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.2.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.2.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.2.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.3.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.3.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.3.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.3.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.3.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.2.self_attns.3.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.input_layernorms.0.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.input_layernorms.1.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.input_layernorms.2.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.input_layernorms.3.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.mlp.experts.down_proj": "model-00001-of-00002.safetensors", |
| "model.layers.3.mlp.experts.gate_proj": "model-00001-of-00002.safetensors", |
| "model.layers.3.mlp.experts.up_proj": "model-00001-of-00002.safetensors", |
| "model.layers.3.mlp.gate.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.post_attention_layernorms.0.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.post_attention_layernorms.1.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.post_attention_layernorms.2.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.post_attention_layernorms.3.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.0.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.0.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.0.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.0.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.0.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.0.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.1.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.1.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.1.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.1.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.1.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.1.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.2.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.2.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.2.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.2.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.2.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.2.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.3.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.3.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.3.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.3.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.3.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.3.self_attns.3.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.input_layernorms.0.weight": "model-00002-of-00002.safetensors", |
| "model.layers.4.input_layernorms.1.weight": "model-00002-of-00002.safetensors", |
| "model.layers.4.input_layernorms.2.weight": "model-00002-of-00002.safetensors", |
| "model.layers.4.input_layernorms.3.weight": "model-00002-of-00002.safetensors", |
| "model.layers.4.mlp.experts.down_proj": "model-00002-of-00002.safetensors", |
| "model.layers.4.mlp.experts.gate_proj": "model-00002-of-00002.safetensors", |
| "model.layers.4.mlp.experts.up_proj": "model-00002-of-00002.safetensors", |
| "model.layers.4.mlp.gate.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.post_attention_layernorms.0.weight": "model-00002-of-00002.safetensors", |
| "model.layers.4.post_attention_layernorms.1.weight": "model-00002-of-00002.safetensors", |
| "model.layers.4.post_attention_layernorms.2.weight": "model-00002-of-00002.safetensors", |
| "model.layers.4.post_attention_layernorms.3.weight": "model-00002-of-00002.safetensors", |
| "model.layers.4.self_attns.0.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.0.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.0.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.0.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.0.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.0.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.1.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.1.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.1.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.1.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.1.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.1.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.2.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.2.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.2.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.2.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.2.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.2.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.3.k_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.3.k_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.3.o_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.3.q_norm.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.3.q_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.4.self_attns.3.v_proj.weight": "model-00001-of-00002.safetensors", |
| "model.layers.5.input_layernorms.0.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.input_layernorms.1.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.input_layernorms.2.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.input_layernorms.3.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.mlp.experts.down_proj": "model-00002-of-00002.safetensors", |
| "model.layers.5.mlp.experts.gate_proj": "model-00002-of-00002.safetensors", |
| "model.layers.5.mlp.experts.up_proj": "model-00002-of-00002.safetensors", |
| "model.layers.5.mlp.gate.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.post_attention_layernorms.0.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.post_attention_layernorms.1.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.post_attention_layernorms.2.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.post_attention_layernorms.3.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.0.k_norm.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.0.k_proj.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.0.o_proj.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.0.q_norm.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.0.q_proj.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.0.v_proj.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.1.k_norm.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.1.k_proj.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.1.o_proj.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.1.q_norm.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.1.q_proj.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.1.v_proj.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.2.k_norm.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.2.k_proj.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.2.o_proj.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.2.q_norm.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.2.q_proj.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.2.v_proj.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.3.k_norm.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.3.k_proj.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.3.o_proj.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.3.q_norm.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.3.q_proj.weight": "model-00002-of-00002.safetensors", |
| "model.layers.5.self_attns.3.v_proj.weight": "model-00002-of-00002.safetensors", |
| "model.norm.weight": "model-00002-of-00002.safetensors" |
| } |
| } |
|
|
