qwen_agent_world_35b_a3b / model.weights.json
laxmareddyp's picture
Upload folder using huggingface_hub
62c7d34 verified
Raw
History Blame Contribute Delete
67.8 kB
{
"metadata": {
"total_size": 138642442752.0
},
"weight_map": {
"/layers/reversible_embedding/vars": [
"model_00000.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder/mlp/expert_bank/vars": [
"model_00000.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder/mlp/shared_expert/_feedforward_output_dense/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder/mlp/_router_gate/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder/mlp/_shared_expert_gate/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder/_input_layernorm/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder/_linear_attn/vars": [
"model_00000.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder/_linear_attn/in_proj_a/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder/_linear_attn/in_proj_b/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder/_linear_attn/in_proj_qkv/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder/_linear_attn/in_proj_z/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder/_linear_attn/norm/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder/_linear_attn/out_proj/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder/_post_attention_layernorm/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_1/mlp/expert_bank/vars": [
"model_00000.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_1/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_1/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_1/mlp/shared_expert/_feedforward_output_dense/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_1/mlp/_router_gate/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_1/mlp/_shared_expert_gate/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_1/_input_layernorm/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_1/_linear_attn/vars": [
"model_00000.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_1/_linear_attn/in_proj_a/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_1/_linear_attn/in_proj_b/vars": "model_00000.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_1/_linear_attn/in_proj_qkv/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_1/_linear_attn/in_proj_z/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_1/_linear_attn/norm/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_1/_linear_attn/out_proj/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_1/_post_attention_layernorm/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_2/mlp/expert_bank/vars": [
"model_00001.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_2/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_2/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_2/mlp/shared_expert/_feedforward_output_dense/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_2/mlp/_router_gate/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_2/mlp/_shared_expert_gate/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_2/_input_layernorm/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_2/_linear_attn/vars": [
"model_00001.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_2/_linear_attn/in_proj_a/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_2/_linear_attn/in_proj_b/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_2/_linear_attn/in_proj_qkv/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_2/_linear_attn/in_proj_z/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_2/_linear_attn/norm/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_2/_linear_attn/out_proj/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_2/_post_attention_layernorm/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_3/mlp/expert_bank/vars": [
"model_00001.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_3/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_3/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_3/mlp/shared_expert/_feedforward_output_dense/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_3/mlp/_router_gate/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_3/mlp/_shared_expert_gate/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_3/_input_layernorm/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_3/_post_attention_layernorm/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_3/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_3/_self_attention_layer/_key_norm/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_3/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_3/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_3/_self_attention_layer/_query_norm/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_3/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_4/mlp/expert_bank/vars": [
"model_00001.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_4/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_4/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_4/mlp/shared_expert/_feedforward_output_dense/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_4/mlp/_router_gate/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_4/mlp/_shared_expert_gate/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_4/_input_layernorm/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_4/_linear_attn/vars": [
"model_00001.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_4/_linear_attn/in_proj_a/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_4/_linear_attn/in_proj_b/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_4/_linear_attn/in_proj_qkv/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_4/_linear_attn/in_proj_z/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_4/_linear_attn/norm/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_4/_linear_attn/out_proj/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_4/_post_attention_layernorm/vars": "model_00001.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_5/mlp/expert_bank/vars": [
"model_00002.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_5/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_5/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_5/mlp/shared_expert/_feedforward_output_dense/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_5/mlp/_router_gate/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_5/mlp/_shared_expert_gate/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_5/_input_layernorm/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_5/_linear_attn/vars": [
"model_00002.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_5/_linear_attn/in_proj_a/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_5/_linear_attn/in_proj_b/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_5/_linear_attn/in_proj_qkv/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_5/_linear_attn/in_proj_z/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_5/_linear_attn/norm/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_5/_linear_attn/out_proj/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_5/_post_attention_layernorm/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_6/mlp/expert_bank/vars": [
"model_00002.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_6/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_6/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_6/mlp/shared_expert/_feedforward_output_dense/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_6/mlp/_router_gate/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_6/mlp/_shared_expert_gate/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_6/_input_layernorm/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_6/_linear_attn/vars": [
"model_00002.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_6/_linear_attn/in_proj_a/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_6/_linear_attn/in_proj_b/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_6/_linear_attn/in_proj_qkv/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_6/_linear_attn/in_proj_z/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_6/_linear_attn/norm/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_6/_linear_attn/out_proj/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_6/_post_attention_layernorm/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_7/mlp/expert_bank/vars": [
"model_00002.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_7/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_7/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_7/mlp/shared_expert/_feedforward_output_dense/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_7/mlp/_router_gate/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_7/mlp/_shared_expert_gate/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_7/_input_layernorm/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_7/_post_attention_layernorm/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_7/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_7/_self_attention_layer/_key_norm/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_7/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_7/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_7/_self_attention_layer/_query_norm/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_7/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_8/mlp/expert_bank/vars": [
"model_00003.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_8/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_8/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_8/mlp/shared_expert/_feedforward_output_dense/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_8/mlp/_router_gate/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_8/mlp/_shared_expert_gate/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_8/_input_layernorm/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_8/_linear_attn/vars": [
"model_00003.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_8/_linear_attn/in_proj_a/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_8/_linear_attn/in_proj_b/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_8/_linear_attn/in_proj_qkv/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_8/_linear_attn/in_proj_z/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_8/_linear_attn/norm/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_8/_linear_attn/out_proj/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_8/_post_attention_layernorm/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_9/mlp/expert_bank/vars": [
"model_00003.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_9/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_9/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_9/mlp/shared_expert/_feedforward_output_dense/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_9/mlp/_router_gate/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_9/mlp/_shared_expert_gate/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_9/_input_layernorm/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_9/_linear_attn/vars": [
"model_00003.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_9/_linear_attn/in_proj_a/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_9/_linear_attn/in_proj_b/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_9/_linear_attn/in_proj_qkv/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_9/_linear_attn/in_proj_z/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_9/_linear_attn/norm/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_9/_linear_attn/out_proj/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_9/_post_attention_layernorm/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_10/mlp/expert_bank/vars": [
"model_00003.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_10/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_10/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_10/mlp/shared_expert/_feedforward_output_dense/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_10/mlp/_router_gate/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_10/mlp/_shared_expert_gate/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_10/_input_layernorm/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_10/_linear_attn/vars": [
"model_00003.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_10/_linear_attn/in_proj_a/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_10/_linear_attn/in_proj_b/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_10/_linear_attn/in_proj_qkv/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_10/_linear_attn/in_proj_z/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_10/_linear_attn/norm/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_10/_linear_attn/out_proj/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_10/_post_attention_layernorm/vars": "model_00003.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_11/mlp/expert_bank/vars": [
"model_00004.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_11/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_11/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_11/mlp/shared_expert/_feedforward_output_dense/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_11/mlp/_router_gate/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_11/mlp/_shared_expert_gate/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_11/_input_layernorm/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_11/_post_attention_layernorm/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_11/_self_attention_layer/_key_dense/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_11/_self_attention_layer/_key_norm/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_11/_self_attention_layer/_output_dense/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_11/_self_attention_layer/_query_dense/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_11/_self_attention_layer/_query_norm/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_11/_self_attention_layer/_value_dense/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_12/mlp/expert_bank/vars": [
"model_00004.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_12/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_12/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_12/mlp/shared_expert/_feedforward_output_dense/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_12/mlp/_router_gate/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_12/mlp/_shared_expert_gate/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_12/_input_layernorm/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_12/_linear_attn/vars": [
"model_00004.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_12/_linear_attn/in_proj_a/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_12/_linear_attn/in_proj_b/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_12/_linear_attn/in_proj_qkv/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_12/_linear_attn/in_proj_z/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_12/_linear_attn/norm/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_12/_linear_attn/out_proj/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_12/_post_attention_layernorm/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_13/mlp/expert_bank/vars": [
"model_00004.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_13/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_13/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_13/mlp/shared_expert/_feedforward_output_dense/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_13/mlp/_router_gate/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_13/mlp/_shared_expert_gate/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_13/_input_layernorm/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_13/_linear_attn/vars": [
"model_00004.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_13/_linear_attn/in_proj_a/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_13/_linear_attn/in_proj_b/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_13/_linear_attn/in_proj_qkv/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_13/_linear_attn/in_proj_z/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_13/_linear_attn/norm/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_13/_linear_attn/out_proj/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_13/_post_attention_layernorm/vars": "model_00004.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_14/mlp/expert_bank/vars": [
"model_00005.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_14/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_14/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_14/mlp/shared_expert/_feedforward_output_dense/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_14/mlp/_router_gate/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_14/mlp/_shared_expert_gate/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_14/_input_layernorm/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_14/_linear_attn/vars": [
"model_00005.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_14/_linear_attn/in_proj_a/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_14/_linear_attn/in_proj_b/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_14/_linear_attn/in_proj_qkv/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_14/_linear_attn/in_proj_z/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_14/_linear_attn/norm/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_14/_linear_attn/out_proj/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_14/_post_attention_layernorm/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_15/mlp/expert_bank/vars": [
"model_00005.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_15/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_15/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_15/mlp/shared_expert/_feedforward_output_dense/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_15/mlp/_router_gate/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_15/mlp/_shared_expert_gate/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_15/_input_layernorm/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_15/_post_attention_layernorm/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_15/_self_attention_layer/_key_dense/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_15/_self_attention_layer/_key_norm/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_15/_self_attention_layer/_output_dense/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_15/_self_attention_layer/_query_dense/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_15/_self_attention_layer/_query_norm/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_15/_self_attention_layer/_value_dense/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_16/mlp/expert_bank/vars": [
"model_00005.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_16/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_16/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_16/mlp/shared_expert/_feedforward_output_dense/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_16/mlp/_router_gate/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_16/mlp/_shared_expert_gate/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_16/_input_layernorm/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_16/_linear_attn/vars": [
"model_00005.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_16/_linear_attn/in_proj_a/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_16/_linear_attn/in_proj_b/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_16/_linear_attn/in_proj_qkv/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_16/_linear_attn/in_proj_z/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_16/_linear_attn/norm/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_16/_linear_attn/out_proj/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_16/_post_attention_layernorm/vars": "model_00005.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_17/mlp/expert_bank/vars": [
"model_00006.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_17/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_17/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_17/mlp/shared_expert/_feedforward_output_dense/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_17/mlp/_router_gate/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_17/mlp/_shared_expert_gate/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_17/_input_layernorm/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_17/_linear_attn/vars": [
"model_00006.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_17/_linear_attn/in_proj_a/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_17/_linear_attn/in_proj_b/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_17/_linear_attn/in_proj_qkv/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_17/_linear_attn/in_proj_z/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_17/_linear_attn/norm/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_17/_linear_attn/out_proj/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_17/_post_attention_layernorm/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_18/mlp/expert_bank/vars": [
"model_00006.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_18/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_18/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_18/mlp/shared_expert/_feedforward_output_dense/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_18/mlp/_router_gate/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_18/mlp/_shared_expert_gate/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_18/_input_layernorm/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_18/_linear_attn/vars": [
"model_00006.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_18/_linear_attn/in_proj_a/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_18/_linear_attn/in_proj_b/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_18/_linear_attn/in_proj_qkv/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_18/_linear_attn/in_proj_z/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_18/_linear_attn/norm/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_18/_linear_attn/out_proj/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_18/_post_attention_layernorm/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_19/mlp/expert_bank/vars": [
"model_00006.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_19/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_19/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_19/mlp/shared_expert/_feedforward_output_dense/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_19/mlp/_router_gate/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_19/mlp/_shared_expert_gate/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_19/_input_layernorm/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_19/_post_attention_layernorm/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_19/_self_attention_layer/_key_dense/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_19/_self_attention_layer/_key_norm/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_19/_self_attention_layer/_output_dense/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_19/_self_attention_layer/_query_dense/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_19/_self_attention_layer/_query_norm/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_19/_self_attention_layer/_value_dense/vars": "model_00006.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_20/mlp/expert_bank/vars": [
"model_00007.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_20/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_20/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_20/mlp/shared_expert/_feedforward_output_dense/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_20/mlp/_router_gate/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_20/mlp/_shared_expert_gate/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_20/_input_layernorm/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_20/_linear_attn/vars": [
"model_00007.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_20/_linear_attn/in_proj_a/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_20/_linear_attn/in_proj_b/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_20/_linear_attn/in_proj_qkv/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_20/_linear_attn/in_proj_z/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_20/_linear_attn/norm/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_20/_linear_attn/out_proj/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_20/_post_attention_layernorm/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_21/mlp/expert_bank/vars": [
"model_00007.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_21/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_21/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_21/mlp/shared_expert/_feedforward_output_dense/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_21/mlp/_router_gate/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_21/mlp/_shared_expert_gate/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_21/_input_layernorm/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_21/_linear_attn/vars": [
"model_00007.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_21/_linear_attn/in_proj_a/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_21/_linear_attn/in_proj_b/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_21/_linear_attn/in_proj_qkv/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_21/_linear_attn/in_proj_z/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_21/_linear_attn/norm/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_21/_linear_attn/out_proj/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_21/_post_attention_layernorm/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_22/mlp/expert_bank/vars": [
"model_00007.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_22/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_22/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_22/mlp/shared_expert/_feedforward_output_dense/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_22/mlp/_router_gate/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_22/mlp/_shared_expert_gate/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_22/_input_layernorm/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_22/_linear_attn/vars": [
"model_00007.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_22/_linear_attn/in_proj_a/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_22/_linear_attn/in_proj_b/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_22/_linear_attn/in_proj_qkv/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_22/_linear_attn/in_proj_z/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_22/_linear_attn/norm/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_22/_linear_attn/out_proj/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_22/_post_attention_layernorm/vars": "model_00007.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_23/mlp/expert_bank/vars": [
"model_00008.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_23/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_23/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_23/mlp/shared_expert/_feedforward_output_dense/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_23/mlp/_router_gate/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_23/mlp/_shared_expert_gate/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_23/_input_layernorm/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_23/_post_attention_layernorm/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_23/_self_attention_layer/_key_dense/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_23/_self_attention_layer/_key_norm/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_23/_self_attention_layer/_output_dense/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_23/_self_attention_layer/_query_dense/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_23/_self_attention_layer/_query_norm/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_23/_self_attention_layer/_value_dense/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_24/mlp/expert_bank/vars": [
"model_00008.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_24/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_24/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_24/mlp/shared_expert/_feedforward_output_dense/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_24/mlp/_router_gate/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_24/mlp/_shared_expert_gate/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_24/_input_layernorm/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_24/_linear_attn/vars": [
"model_00008.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_24/_linear_attn/in_proj_a/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_24/_linear_attn/in_proj_b/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_24/_linear_attn/in_proj_qkv/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_24/_linear_attn/in_proj_z/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_24/_linear_attn/norm/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_24/_linear_attn/out_proj/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_24/_post_attention_layernorm/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_25/mlp/expert_bank/vars": [
"model_00008.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_25/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_25/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_25/mlp/shared_expert/_feedforward_output_dense/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_25/mlp/_router_gate/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_25/mlp/_shared_expert_gate/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_25/_input_layernorm/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_25/_linear_attn/vars": [
"model_00008.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_25/_linear_attn/in_proj_a/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_25/_linear_attn/in_proj_b/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_25/_linear_attn/in_proj_qkv/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_25/_linear_attn/in_proj_z/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_25/_linear_attn/norm/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_25/_linear_attn/out_proj/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_25/_post_attention_layernorm/vars": "model_00008.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_26/mlp/expert_bank/vars": [
"model_00009.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_26/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_26/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_26/mlp/shared_expert/_feedforward_output_dense/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_26/mlp/_router_gate/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_26/mlp/_shared_expert_gate/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_26/_input_layernorm/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_26/_linear_attn/vars": [
"model_00009.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_26/_linear_attn/in_proj_a/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_26/_linear_attn/in_proj_b/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_26/_linear_attn/in_proj_qkv/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_26/_linear_attn/in_proj_z/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_26/_linear_attn/norm/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_26/_linear_attn/out_proj/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_26/_post_attention_layernorm/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_27/mlp/expert_bank/vars": [
"model_00009.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_27/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_27/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_27/mlp/shared_expert/_feedforward_output_dense/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_27/mlp/_router_gate/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_27/mlp/_shared_expert_gate/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_27/_input_layernorm/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_27/_post_attention_layernorm/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_27/_self_attention_layer/_key_dense/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_27/_self_attention_layer/_key_norm/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_27/_self_attention_layer/_output_dense/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_27/_self_attention_layer/_query_dense/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_27/_self_attention_layer/_query_norm/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_27/_self_attention_layer/_value_dense/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_28/mlp/expert_bank/vars": [
"model_00009.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_28/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_28/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_28/mlp/shared_expert/_feedforward_output_dense/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_28/mlp/_router_gate/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_28/mlp/_shared_expert_gate/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_28/_input_layernorm/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_28/_linear_attn/vars": [
"model_00009.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_28/_linear_attn/in_proj_a/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_28/_linear_attn/in_proj_b/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_28/_linear_attn/in_proj_qkv/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_28/_linear_attn/in_proj_z/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_28/_linear_attn/norm/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_28/_linear_attn/out_proj/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_28/_post_attention_layernorm/vars": "model_00009.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_29/mlp/expert_bank/vars": [
"model_00010.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_29/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_29/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_29/mlp/shared_expert/_feedforward_output_dense/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_29/mlp/_router_gate/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_29/mlp/_shared_expert_gate/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_29/_input_layernorm/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_29/_linear_attn/vars": [
"model_00010.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_29/_linear_attn/in_proj_a/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_29/_linear_attn/in_proj_b/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_29/_linear_attn/in_proj_qkv/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_29/_linear_attn/in_proj_z/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_29/_linear_attn/norm/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_29/_linear_attn/out_proj/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_29/_post_attention_layernorm/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_30/mlp/expert_bank/vars": [
"model_00010.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_30/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_30/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_30/mlp/shared_expert/_feedforward_output_dense/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_30/mlp/_router_gate/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_30/mlp/_shared_expert_gate/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_30/_input_layernorm/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_30/_linear_attn/vars": [
"model_00010.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_30/_linear_attn/in_proj_a/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_30/_linear_attn/in_proj_b/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_30/_linear_attn/in_proj_qkv/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_30/_linear_attn/in_proj_z/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_30/_linear_attn/norm/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_30/_linear_attn/out_proj/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_30/_post_attention_layernorm/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_31/mlp/expert_bank/vars": [
"model_00010.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_31/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_31/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_31/mlp/shared_expert/_feedforward_output_dense/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_31/mlp/_router_gate/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_31/mlp/_shared_expert_gate/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_31/_input_layernorm/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_31/_post_attention_layernorm/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_31/_self_attention_layer/_key_dense/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_31/_self_attention_layer/_key_norm/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_31/_self_attention_layer/_output_dense/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_31/_self_attention_layer/_query_dense/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_31/_self_attention_layer/_query_norm/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_31/_self_attention_layer/_value_dense/vars": "model_00010.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_32/mlp/expert_bank/vars": [
"model_00011.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_32/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_32/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_32/mlp/shared_expert/_feedforward_output_dense/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_32/mlp/_router_gate/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_32/mlp/_shared_expert_gate/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_32/_input_layernorm/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_32/_linear_attn/vars": [
"model_00011.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_32/_linear_attn/in_proj_a/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_32/_linear_attn/in_proj_b/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_32/_linear_attn/in_proj_qkv/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_32/_linear_attn/in_proj_z/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_32/_linear_attn/norm/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_32/_linear_attn/out_proj/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_32/_post_attention_layernorm/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_33/mlp/expert_bank/vars": [
"model_00011.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_33/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_33/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_33/mlp/shared_expert/_feedforward_output_dense/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_33/mlp/_router_gate/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_33/mlp/_shared_expert_gate/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_33/_input_layernorm/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_33/_linear_attn/vars": [
"model_00011.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_33/_linear_attn/in_proj_a/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_33/_linear_attn/in_proj_b/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_33/_linear_attn/in_proj_qkv/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_33/_linear_attn/in_proj_z/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_33/_linear_attn/norm/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_33/_linear_attn/out_proj/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_33/_post_attention_layernorm/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_34/mlp/expert_bank/vars": [
"model_00011.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_34/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_34/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_34/mlp/shared_expert/_feedforward_output_dense/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_34/mlp/_router_gate/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_34/mlp/_shared_expert_gate/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_34/_input_layernorm/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_34/_linear_attn/vars": [
"model_00011.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_34/_linear_attn/in_proj_a/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_34/_linear_attn/in_proj_b/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_34/_linear_attn/in_proj_qkv/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_34/_linear_attn/in_proj_z/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_34/_linear_attn/norm/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_34/_linear_attn/out_proj/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_34/_post_attention_layernorm/vars": "model_00011.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_35/mlp/expert_bank/vars": [
"model_00012.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_35/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_35/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_35/mlp/shared_expert/_feedforward_output_dense/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_35/mlp/_router_gate/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_35/mlp/_shared_expert_gate/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_35/_input_layernorm/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_35/_post_attention_layernorm/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_35/_self_attention_layer/_key_dense/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_35/_self_attention_layer/_key_norm/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_35/_self_attention_layer/_output_dense/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_35/_self_attention_layer/_query_dense/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_35/_self_attention_layer/_query_norm/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_35/_self_attention_layer/_value_dense/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_36/mlp/expert_bank/vars": [
"model_00012.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_36/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_36/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_36/mlp/shared_expert/_feedforward_output_dense/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_36/mlp/_router_gate/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_36/mlp/_shared_expert_gate/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_36/_input_layernorm/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_36/_linear_attn/vars": [
"model_00012.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_36/_linear_attn/in_proj_a/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_36/_linear_attn/in_proj_b/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_36/_linear_attn/in_proj_qkv/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_36/_linear_attn/in_proj_z/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_36/_linear_attn/norm/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_36/_linear_attn/out_proj/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_36/_post_attention_layernorm/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_37/mlp/expert_bank/vars": [
"model_00012.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_37/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_37/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_37/mlp/shared_expert/_feedforward_output_dense/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_37/mlp/_router_gate/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_37/mlp/_shared_expert_gate/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_37/_input_layernorm/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_37/_linear_attn/vars": [
"model_00012.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_37/_linear_attn/in_proj_a/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_37/_linear_attn/in_proj_b/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_37/_linear_attn/in_proj_qkv/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_37/_linear_attn/in_proj_z/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_37/_linear_attn/norm/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_37/_linear_attn/out_proj/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_37/_post_attention_layernorm/vars": "model_00012.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_38/mlp/expert_bank/vars": [
"model_00013.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_38/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_38/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_38/mlp/shared_expert/_feedforward_output_dense/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_38/mlp/_router_gate/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_38/mlp/_shared_expert_gate/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_38/_input_layernorm/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_38/_linear_attn/vars": [
"model_00013.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_38/_linear_attn/in_proj_a/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_38/_linear_attn/in_proj_b/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_38/_linear_attn/in_proj_qkv/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_38/_linear_attn/in_proj_z/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_38/_linear_attn/norm/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_38/_linear_attn/out_proj/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_38/_post_attention_layernorm/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_39/mlp/expert_bank/vars": [
"model_00013.weights.h5"
],
"/layers/qwen3_5_moe_transformer_decoder_39/mlp/shared_expert/_feedforward_gate_dense/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_39/mlp/shared_expert/_feedforward_intermediate_dense/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_39/mlp/shared_expert/_feedforward_output_dense/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_39/mlp/_router_gate/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_39/mlp/_shared_expert_gate/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_39/_input_layernorm/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_39/_post_attention_layernorm/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_39/_self_attention_layer/_key_dense/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_39/_self_attention_layer/_key_norm/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_39/_self_attention_layer/_output_dense/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_39/_self_attention_layer/_query_dense/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_39/_self_attention_layer/_query_norm/vars": "model_00013.weights.h5",
"/layers/qwen3_5_moe_transformer_decoder_39/_self_attention_layer/_value_dense/vars": "model_00013.weights.h5",
"/layers/qwen3_5_layer_norm/vars": "model_00013.weights.h5"
}
}