| { |
| "metadata": { |
| "total_size": 213704502528.0 |
| }, |
| "weight_map": { |
| "/layers/embedding/vars": [ |
| "model_00000.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer/input_layernorm/vars": [ |
| "model_00000.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer/mlp/down/vars": [ |
| "model_00000.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer/mlp/gate/vars": [ |
| "model_00000.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer/mlp/up/vars": [ |
| "model_00000.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer/post_attention_layernorm/vars": [ |
| "model_00000.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer/attention/key/vars": [ |
| "model_00000.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer/attention/output_proj/vars": [ |
| "model_00000.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer/attention/query/vars": [ |
| "model_00000.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer/attention/value/vars": [ |
| "model_00000.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_1/input_layernorm/vars": [ |
| "model_00000.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_1/mlp/experts/vars": [ |
| "model_00000.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_1/mlp/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_1/mlp/shared_experts/down/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_1/mlp/shared_experts/gate/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_1/mlp/shared_experts/up/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_1/post_attention_layernorm/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_1/attention/key/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_1/attention/output_proj/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_1/attention/query/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_1/attention/value/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_2/input_layernorm/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_2/mlp/experts/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_2/mlp/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_2/mlp/shared_experts/down/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_2/mlp/shared_experts/gate/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_2/mlp/shared_experts/up/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_2/post_attention_layernorm/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_2/attention/key/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_2/attention/output_proj/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_2/attention/query/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_2/attention/value/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_3/input_layernorm/vars": [ |
| "model_00001.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_3/mlp/experts/vars": [ |
| "model_00002.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_3/mlp/vars": [ |
| "model_00002.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_3/mlp/shared_experts/down/vars": [ |
| "model_00002.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_3/mlp/shared_experts/gate/vars": [ |
| "model_00002.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_3/mlp/shared_experts/up/vars": [ |
| "model_00002.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_3/post_attention_layernorm/vars": [ |
| "model_00002.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_3/attention/key/vars": [ |
| "model_00002.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_3/attention/output_proj/vars": [ |
| "model_00002.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_3/attention/query/vars": [ |
| "model_00002.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_3/attention/value/vars": [ |
| "model_00002.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_4/input_layernorm/vars": [ |
| "model_00002.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_4/mlp/experts/vars": [ |
| "model_00003.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_4/mlp/vars": [ |
| "model_00003.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_4/mlp/shared_experts/down/vars": [ |
| "model_00003.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_4/mlp/shared_experts/gate/vars": [ |
| "model_00003.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_4/mlp/shared_experts/up/vars": [ |
| "model_00003.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_4/post_attention_layernorm/vars": [ |
| "model_00003.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_4/attention/key/vars": [ |
| "model_00003.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_4/attention/output_proj/vars": [ |
| "model_00003.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_4/attention/query/vars": [ |
| "model_00003.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_4/attention/value/vars": [ |
| "model_00003.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_5/input_layernorm/vars": [ |
| "model_00003.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_5/mlp/experts/vars": [ |
| "model_00004.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_5/mlp/vars": [ |
| "model_00004.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_5/mlp/shared_experts/down/vars": [ |
| "model_00004.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_5/mlp/shared_experts/gate/vars": [ |
| "model_00004.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_5/mlp/shared_experts/up/vars": [ |
| "model_00004.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_5/post_attention_layernorm/vars": [ |
| "model_00004.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_5/attention/key/vars": [ |
| "model_00004.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_5/attention/output_proj/vars": [ |
| "model_00004.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_5/attention/query/vars": [ |
| "model_00004.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_5/attention/value/vars": [ |
| "model_00004.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_6/input_layernorm/vars": [ |
| "model_00004.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_6/mlp/experts/vars": [ |
| "model_00005.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_6/mlp/vars": [ |
| "model_00005.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_6/mlp/shared_experts/down/vars": [ |
| "model_00005.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_6/mlp/shared_experts/gate/vars": [ |
| "model_00005.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_6/mlp/shared_experts/up/vars": [ |
| "model_00005.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_6/post_attention_layernorm/vars": [ |
| "model_00005.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_6/attention/key/vars": [ |
| "model_00005.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_6/attention/output_proj/vars": [ |
| "model_00005.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_6/attention/query/vars": [ |
| "model_00005.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_6/attention/value/vars": [ |
| "model_00005.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_7/input_layernorm/vars": [ |
| "model_00005.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_7/mlp/experts/vars": [ |
| "model_00006.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_7/mlp/vars": [ |
| "model_00006.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_7/mlp/shared_experts/down/vars": [ |
| "model_00006.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_7/mlp/shared_experts/gate/vars": [ |
| "model_00006.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_7/mlp/shared_experts/up/vars": [ |
| "model_00006.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_7/post_attention_layernorm/vars": [ |
| "model_00006.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_7/attention/key/vars": [ |
| "model_00006.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_7/attention/output_proj/vars": [ |
| "model_00006.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_7/attention/query/vars": [ |
| "model_00006.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_7/attention/value/vars": [ |
| "model_00006.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_8/input_layernorm/vars": [ |
| "model_00006.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_8/mlp/experts/vars": [ |
| "model_00007.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_8/mlp/vars": [ |
| "model_00007.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_8/mlp/shared_experts/down/vars": [ |
| "model_00007.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_8/mlp/shared_experts/gate/vars": [ |
| "model_00007.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_8/mlp/shared_experts/up/vars": [ |
| "model_00007.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_8/post_attention_layernorm/vars": [ |
| "model_00007.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_8/attention/key/vars": [ |
| "model_00007.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_8/attention/output_proj/vars": [ |
| "model_00007.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_8/attention/query/vars": [ |
| "model_00007.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_8/attention/value/vars": [ |
| "model_00007.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_9/input_layernorm/vars": [ |
| "model_00007.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_9/mlp/experts/vars": [ |
| "model_00008.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_9/mlp/vars": [ |
| "model_00008.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_9/mlp/shared_experts/down/vars": [ |
| "model_00008.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_9/mlp/shared_experts/gate/vars": [ |
| "model_00008.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_9/mlp/shared_experts/up/vars": [ |
| "model_00008.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_9/post_attention_layernorm/vars": [ |
| "model_00008.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_9/attention/key/vars": [ |
| "model_00008.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_9/attention/output_proj/vars": [ |
| "model_00008.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_9/attention/query/vars": [ |
| "model_00008.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_9/attention/value/vars": [ |
| "model_00008.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_10/input_layernorm/vars": [ |
| "model_00008.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_10/mlp/experts/vars": [ |
| "model_00009.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_10/mlp/vars": [ |
| "model_00009.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_10/mlp/shared_experts/down/vars": [ |
| "model_00009.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_10/mlp/shared_experts/gate/vars": [ |
| "model_00009.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_10/mlp/shared_experts/up/vars": [ |
| "model_00009.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_10/post_attention_layernorm/vars": [ |
| "model_00009.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_10/attention/key/vars": [ |
| "model_00009.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_10/attention/output_proj/vars": [ |
| "model_00009.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_10/attention/query/vars": [ |
| "model_00009.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_10/attention/value/vars": [ |
| "model_00009.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_11/input_layernorm/vars": [ |
| "model_00009.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_11/mlp/experts/vars": [ |
| "model_00010.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_11/mlp/vars": [ |
| "model_00010.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_11/mlp/shared_experts/down/vars": [ |
| "model_00010.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_11/mlp/shared_experts/gate/vars": [ |
| "model_00010.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_11/mlp/shared_experts/up/vars": [ |
| "model_00010.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_11/post_attention_layernorm/vars": [ |
| "model_00010.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_11/attention/key/vars": [ |
| "model_00010.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_11/attention/output_proj/vars": [ |
| "model_00010.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_11/attention/query/vars": [ |
| "model_00010.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_11/attention/value/vars": [ |
| "model_00010.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_12/input_layernorm/vars": [ |
| "model_00010.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_12/mlp/experts/vars": [ |
| "model_00011.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_12/mlp/vars": [ |
| "model_00011.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_12/mlp/shared_experts/down/vars": [ |
| "model_00011.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_12/mlp/shared_experts/gate/vars": [ |
| "model_00011.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_12/mlp/shared_experts/up/vars": [ |
| "model_00011.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_12/post_attention_layernorm/vars": [ |
| "model_00011.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_12/attention/key/vars": [ |
| "model_00011.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_12/attention/output_proj/vars": [ |
| "model_00011.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_12/attention/query/vars": [ |
| "model_00011.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_12/attention/value/vars": [ |
| "model_00011.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_13/input_layernorm/vars": [ |
| "model_00011.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_13/mlp/experts/vars": [ |
| "model_00012.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_13/mlp/vars": [ |
| "model_00012.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_13/mlp/shared_experts/down/vars": [ |
| "model_00012.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_13/mlp/shared_experts/gate/vars": [ |
| "model_00012.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_13/mlp/shared_experts/up/vars": [ |
| "model_00012.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_13/post_attention_layernorm/vars": [ |
| "model_00012.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_13/attention/key/vars": [ |
| "model_00012.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_13/attention/output_proj/vars": [ |
| "model_00012.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_13/attention/query/vars": [ |
| "model_00012.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_13/attention/value/vars": [ |
| "model_00012.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_14/input_layernorm/vars": [ |
| "model_00012.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_14/mlp/experts/vars": [ |
| "model_00013.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_14/mlp/vars": [ |
| "model_00013.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_14/mlp/shared_experts/down/vars": [ |
| "model_00013.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_14/mlp/shared_experts/gate/vars": [ |
| "model_00013.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_14/mlp/shared_experts/up/vars": [ |
| "model_00013.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_14/post_attention_layernorm/vars": [ |
| "model_00013.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_14/attention/key/vars": [ |
| "model_00013.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_14/attention/output_proj/vars": [ |
| "model_00013.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_14/attention/query/vars": [ |
| "model_00013.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_14/attention/value/vars": [ |
| "model_00013.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_15/input_layernorm/vars": [ |
| "model_00013.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_15/mlp/experts/vars": [ |
| "model_00014.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_15/mlp/vars": [ |
| "model_00014.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_15/mlp/shared_experts/down/vars": [ |
| "model_00014.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_15/mlp/shared_experts/gate/vars": [ |
| "model_00014.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_15/mlp/shared_experts/up/vars": [ |
| "model_00014.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_15/post_attention_layernorm/vars": [ |
| "model_00014.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_15/attention/key/vars": [ |
| "model_00014.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_15/attention/output_proj/vars": [ |
| "model_00014.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_15/attention/query/vars": [ |
| "model_00014.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_15/attention/value/vars": [ |
| "model_00014.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_16/input_layernorm/vars": [ |
| "model_00014.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_16/mlp/experts/vars": [ |
| "model_00015.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_16/mlp/vars": [ |
| "model_00015.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_16/mlp/shared_experts/down/vars": [ |
| "model_00015.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_16/mlp/shared_experts/gate/vars": [ |
| "model_00015.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_16/mlp/shared_experts/up/vars": [ |
| "model_00015.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_16/post_attention_layernorm/vars": [ |
| "model_00015.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_16/attention/key/vars": [ |
| "model_00015.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_16/attention/output_proj/vars": [ |
| "model_00015.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_16/attention/query/vars": [ |
| "model_00015.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_16/attention/value/vars": [ |
| "model_00015.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_17/input_layernorm/vars": [ |
| "model_00015.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_17/mlp/experts/vars": [ |
| "model_00016.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_17/mlp/vars": [ |
| "model_00016.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_17/mlp/shared_experts/down/vars": [ |
| "model_00016.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_17/mlp/shared_experts/gate/vars": [ |
| "model_00016.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_17/mlp/shared_experts/up/vars": [ |
| "model_00016.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_17/post_attention_layernorm/vars": [ |
| "model_00016.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_17/attention/key/vars": [ |
| "model_00016.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_17/attention/output_proj/vars": [ |
| "model_00016.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_17/attention/query/vars": [ |
| "model_00016.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_17/attention/value/vars": [ |
| "model_00016.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_18/input_layernorm/vars": [ |
| "model_00016.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_18/mlp/experts/vars": [ |
| "model_00017.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_18/mlp/vars": [ |
| "model_00017.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_18/mlp/shared_experts/down/vars": [ |
| "model_00017.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_18/mlp/shared_experts/gate/vars": [ |
| "model_00017.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_18/mlp/shared_experts/up/vars": [ |
| "model_00017.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_18/post_attention_layernorm/vars": [ |
| "model_00017.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_18/attention/key/vars": [ |
| "model_00017.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_18/attention/output_proj/vars": [ |
| "model_00017.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_18/attention/query/vars": [ |
| "model_00017.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_18/attention/value/vars": [ |
| "model_00017.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_19/input_layernorm/vars": [ |
| "model_00017.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_19/mlp/experts/vars": [ |
| "model_00018.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_19/mlp/vars": [ |
| "model_00018.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_19/mlp/shared_experts/down/vars": [ |
| "model_00018.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_19/mlp/shared_experts/gate/vars": [ |
| "model_00018.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_19/mlp/shared_experts/up/vars": [ |
| "model_00018.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_19/post_attention_layernorm/vars": [ |
| "model_00018.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_19/attention/key/vars": [ |
| "model_00018.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_19/attention/output_proj/vars": [ |
| "model_00018.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_19/attention/query/vars": [ |
| "model_00018.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_19/attention/value/vars": [ |
| "model_00018.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_20/input_layernorm/vars": [ |
| "model_00018.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_20/mlp/experts/vars": [ |
| "model_00019.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_20/mlp/vars": [ |
| "model_00019.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_20/mlp/shared_experts/down/vars": [ |
| "model_00019.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_20/mlp/shared_experts/gate/vars": [ |
| "model_00019.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_20/mlp/shared_experts/up/vars": [ |
| "model_00019.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_20/post_attention_layernorm/vars": [ |
| "model_00019.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_20/attention/key/vars": [ |
| "model_00019.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_20/attention/output_proj/vars": [ |
| "model_00019.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_20/attention/query/vars": [ |
| "model_00019.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_20/attention/value/vars": [ |
| "model_00019.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_21/input_layernorm/vars": [ |
| "model_00019.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_21/mlp/experts/vars": [ |
| "model_00020.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_21/mlp/vars": [ |
| "model_00020.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_21/mlp/shared_experts/down/vars": [ |
| "model_00020.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_21/mlp/shared_experts/gate/vars": [ |
| "model_00020.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_21/mlp/shared_experts/up/vars": [ |
| "model_00020.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_21/post_attention_layernorm/vars": [ |
| "model_00020.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_21/attention/key/vars": [ |
| "model_00020.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_21/attention/output_proj/vars": [ |
| "model_00020.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_21/attention/query/vars": [ |
| "model_00020.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_21/attention/value/vars": [ |
| "model_00020.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_22/input_layernorm/vars": [ |
| "model_00020.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_22/mlp/experts/vars": [ |
| "model_00021.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_22/mlp/vars": [ |
| "model_00021.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_22/mlp/shared_experts/down/vars": [ |
| "model_00021.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_22/mlp/shared_experts/gate/vars": [ |
| "model_00021.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_22/mlp/shared_experts/up/vars": [ |
| "model_00021.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_22/post_attention_layernorm/vars": [ |
| "model_00021.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_22/attention/key/vars": [ |
| "model_00021.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_22/attention/output_proj/vars": [ |
| "model_00021.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_22/attention/query/vars": [ |
| "model_00021.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_22/attention/value/vars": [ |
| "model_00021.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_23/input_layernorm/vars": [ |
| "model_00021.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_23/mlp/experts/vars": [ |
| "model_00022.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_23/mlp/vars": [ |
| "model_00022.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_23/mlp/shared_experts/down/vars": [ |
| "model_00022.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_23/mlp/shared_experts/gate/vars": [ |
| "model_00022.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_23/mlp/shared_experts/up/vars": [ |
| "model_00022.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_23/post_attention_layernorm/vars": [ |
| "model_00022.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_23/attention/key/vars": [ |
| "model_00022.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_23/attention/output_proj/vars": [ |
| "model_00022.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_23/attention/query/vars": [ |
| "model_00022.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_23/attention/value/vars": [ |
| "model_00022.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_24/input_layernorm/vars": [ |
| "model_00022.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_24/mlp/experts/vars": [ |
| "model_00023.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_24/mlp/vars": [ |
| "model_00023.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_24/mlp/shared_experts/down/vars": [ |
| "model_00023.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_24/mlp/shared_experts/gate/vars": [ |
| "model_00023.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_24/mlp/shared_experts/up/vars": [ |
| "model_00023.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_24/post_attention_layernorm/vars": [ |
| "model_00023.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_24/attention/key/vars": [ |
| "model_00023.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_24/attention/output_proj/vars": [ |
| "model_00023.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_24/attention/query/vars": [ |
| "model_00023.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_24/attention/value/vars": [ |
| "model_00023.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_25/input_layernorm/vars": [ |
| "model_00023.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_25/mlp/experts/vars": [ |
| "model_00024.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_25/mlp/vars": [ |
| "model_00024.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_25/mlp/shared_experts/down/vars": [ |
| "model_00024.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_25/mlp/shared_experts/gate/vars": [ |
| "model_00024.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_25/mlp/shared_experts/up/vars": [ |
| "model_00024.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_25/post_attention_layernorm/vars": [ |
| "model_00024.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_25/attention/key/vars": [ |
| "model_00024.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_25/attention/output_proj/vars": [ |
| "model_00024.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_25/attention/query/vars": [ |
| "model_00024.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_25/attention/value/vars": [ |
| "model_00024.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_26/input_layernorm/vars": [ |
| "model_00024.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_26/mlp/experts/vars": [ |
| "model_00025.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_26/mlp/vars": [ |
| "model_00025.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_26/mlp/shared_experts/down/vars": [ |
| "model_00025.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_26/mlp/shared_experts/gate/vars": [ |
| "model_00025.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_26/mlp/shared_experts/up/vars": [ |
| "model_00025.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_26/post_attention_layernorm/vars": [ |
| "model_00025.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_26/attention/key/vars": [ |
| "model_00025.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_26/attention/output_proj/vars": [ |
| "model_00025.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_26/attention/query/vars": [ |
| "model_00025.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_26/attention/value/vars": [ |
| "model_00025.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_27/input_layernorm/vars": [ |
| "model_00025.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_27/mlp/experts/vars": [ |
| "model_00026.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_27/mlp/vars": [ |
| "model_00026.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_27/mlp/shared_experts/down/vars": [ |
| "model_00026.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_27/mlp/shared_experts/gate/vars": [ |
| "model_00026.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_27/mlp/shared_experts/up/vars": [ |
| "model_00026.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_27/post_attention_layernorm/vars": [ |
| "model_00026.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_27/attention/key/vars": [ |
| "model_00026.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_27/attention/output_proj/vars": [ |
| "model_00026.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_27/attention/query/vars": [ |
| "model_00026.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_27/attention/value/vars": [ |
| "model_00026.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_28/input_layernorm/vars": [ |
| "model_00026.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_28/mlp/experts/vars": [ |
| "model_00027.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_28/mlp/vars": [ |
| "model_00027.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_28/mlp/shared_experts/down/vars": [ |
| "model_00027.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_28/mlp/shared_experts/gate/vars": [ |
| "model_00027.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_28/mlp/shared_experts/up/vars": [ |
| "model_00027.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_28/post_attention_layernorm/vars": [ |
| "model_00027.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_28/attention/key/vars": [ |
| "model_00027.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_28/attention/output_proj/vars": [ |
| "model_00027.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_28/attention/query/vars": [ |
| "model_00027.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_28/attention/value/vars": [ |
| "model_00027.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_29/input_layernorm/vars": [ |
| "model_00027.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_29/mlp/experts/vars": [ |
| "model_00028.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_29/mlp/vars": [ |
| "model_00028.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_29/mlp/shared_experts/down/vars": [ |
| "model_00028.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_29/mlp/shared_experts/gate/vars": [ |
| "model_00028.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_29/mlp/shared_experts/up/vars": [ |
| "model_00028.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_29/post_attention_layernorm/vars": [ |
| "model_00028.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_29/attention/key/vars": [ |
| "model_00028.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_29/attention/output_proj/vars": [ |
| "model_00028.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_29/attention/query/vars": [ |
| "model_00028.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_29/attention/value/vars": [ |
| "model_00028.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_30/input_layernorm/vars": [ |
| "model_00028.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_30/mlp/experts/vars": [ |
| "model_00029.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_30/mlp/vars": [ |
| "model_00029.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_30/mlp/shared_experts/down/vars": [ |
| "model_00029.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_30/mlp/shared_experts/gate/vars": [ |
| "model_00029.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_30/mlp/shared_experts/up/vars": [ |
| "model_00029.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_30/post_attention_layernorm/vars": [ |
| "model_00029.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_30/attention/key/vars": [ |
| "model_00029.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_30/attention/output_proj/vars": [ |
| "model_00029.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_30/attention/query/vars": [ |
| "model_00029.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_30/attention/value/vars": [ |
| "model_00029.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_31/input_layernorm/vars": [ |
| "model_00029.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_31/mlp/experts/vars": [ |
| "model_00030.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_31/mlp/vars": [ |
| "model_00030.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_31/mlp/shared_experts/down/vars": [ |
| "model_00030.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_31/mlp/shared_experts/gate/vars": [ |
| "model_00030.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_31/mlp/shared_experts/up/vars": [ |
| "model_00030.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_31/post_attention_layernorm/vars": [ |
| "model_00030.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_31/attention/key/vars": [ |
| "model_00030.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_31/attention/output_proj/vars": [ |
| "model_00030.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_31/attention/query/vars": [ |
| "model_00030.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_31/attention/value/vars": [ |
| "model_00030.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_32/input_layernorm/vars": [ |
| "model_00030.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_32/mlp/experts/vars": [ |
| "model_00031.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_32/mlp/vars": [ |
| "model_00031.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_32/mlp/shared_experts/down/vars": [ |
| "model_00031.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_32/mlp/shared_experts/gate/vars": [ |
| "model_00031.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_32/mlp/shared_experts/up/vars": [ |
| "model_00031.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_32/post_attention_layernorm/vars": [ |
| "model_00031.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_32/attention/key/vars": [ |
| "model_00031.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_32/attention/output_proj/vars": [ |
| "model_00031.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_32/attention/query/vars": [ |
| "model_00031.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_32/attention/value/vars": [ |
| "model_00031.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_33/input_layernorm/vars": [ |
| "model_00031.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_33/mlp/experts/vars": [ |
| "model_00032.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_33/mlp/vars": [ |
| "model_00032.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_33/mlp/shared_experts/down/vars": [ |
| "model_00032.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_33/mlp/shared_experts/gate/vars": [ |
| "model_00032.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_33/mlp/shared_experts/up/vars": [ |
| "model_00032.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_33/post_attention_layernorm/vars": [ |
| "model_00032.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_33/attention/key/vars": [ |
| "model_00032.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_33/attention/output_proj/vars": [ |
| "model_00032.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_33/attention/query/vars": [ |
| "model_00032.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_33/attention/value/vars": [ |
| "model_00032.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_34/input_layernorm/vars": [ |
| "model_00032.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_34/mlp/experts/vars": [ |
| "model_00033.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_34/mlp/vars": [ |
| "model_00033.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_34/mlp/shared_experts/down/vars": [ |
| "model_00033.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_34/mlp/shared_experts/gate/vars": [ |
| "model_00033.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_34/mlp/shared_experts/up/vars": [ |
| "model_00033.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_34/post_attention_layernorm/vars": [ |
| "model_00033.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_34/attention/key/vars": [ |
| "model_00033.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_34/attention/output_proj/vars": [ |
| "model_00033.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_34/attention/query/vars": [ |
| "model_00033.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_34/attention/value/vars": [ |
| "model_00033.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_35/input_layernorm/vars": [ |
| "model_00033.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_35/mlp/experts/vars": [ |
| "model_00034.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_35/mlp/vars": [ |
| "model_00034.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_35/mlp/shared_experts/down/vars": [ |
| "model_00034.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_35/mlp/shared_experts/gate/vars": [ |
| "model_00034.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_35/mlp/shared_experts/up/vars": [ |
| "model_00034.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_35/post_attention_layernorm/vars": [ |
| "model_00034.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_35/attention/key/vars": [ |
| "model_00034.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_35/attention/output_proj/vars": [ |
| "model_00034.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_35/attention/query/vars": [ |
| "model_00034.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_35/attention/value/vars": [ |
| "model_00034.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_36/input_layernorm/vars": [ |
| "model_00034.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_36/mlp/experts/vars": [ |
| "model_00035.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_36/mlp/vars": [ |
| "model_00035.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_36/mlp/shared_experts/down/vars": [ |
| "model_00035.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_36/mlp/shared_experts/gate/vars": [ |
| "model_00035.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_36/mlp/shared_experts/up/vars": [ |
| "model_00035.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_36/post_attention_layernorm/vars": [ |
| "model_00035.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_36/attention/key/vars": [ |
| "model_00035.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_36/attention/output_proj/vars": [ |
| "model_00035.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_36/attention/query/vars": [ |
| "model_00035.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_36/attention/value/vars": [ |
| "model_00035.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_37/input_layernorm/vars": [ |
| "model_00035.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_37/mlp/experts/vars": [ |
| "model_00036.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_37/mlp/vars": [ |
| "model_00036.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_37/mlp/shared_experts/down/vars": [ |
| "model_00036.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_37/mlp/shared_experts/gate/vars": [ |
| "model_00036.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_37/mlp/shared_experts/up/vars": [ |
| "model_00036.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_37/post_attention_layernorm/vars": [ |
| "model_00036.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_37/attention/key/vars": [ |
| "model_00036.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_37/attention/output_proj/vars": [ |
| "model_00036.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_37/attention/query/vars": [ |
| "model_00036.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_37/attention/value/vars": [ |
| "model_00036.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_38/input_layernorm/vars": [ |
| "model_00036.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_38/mlp/experts/vars": [ |
| "model_00037.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_38/mlp/vars": [ |
| "model_00037.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_38/mlp/shared_experts/down/vars": [ |
| "model_00037.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_38/mlp/shared_experts/gate/vars": [ |
| "model_00037.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_38/mlp/shared_experts/up/vars": [ |
| "model_00037.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_38/post_attention_layernorm/vars": [ |
| "model_00037.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_38/attention/key/vars": [ |
| "model_00037.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_38/attention/output_proj/vars": [ |
| "model_00037.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_38/attention/query/vars": [ |
| "model_00037.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_38/attention/value/vars": [ |
| "model_00037.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_39/input_layernorm/vars": [ |
| "model_00037.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_39/mlp/experts/vars": [ |
| "model_00038.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_39/mlp/vars": [ |
| "model_00038.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_39/mlp/shared_experts/down/vars": [ |
| "model_00038.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_39/mlp/shared_experts/gate/vars": [ |
| "model_00038.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_39/mlp/shared_experts/up/vars": [ |
| "model_00038.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_39/post_attention_layernorm/vars": [ |
| "model_00038.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_39/attention/key/vars": [ |
| "model_00038.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_39/attention/output_proj/vars": [ |
| "model_00038.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_39/attention/query/vars": [ |
| "model_00038.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_39/attention/value/vars": [ |
| "model_00038.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_40/input_layernorm/vars": [ |
| "model_00038.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_40/mlp/experts/vars": [ |
| "model_00039.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_40/mlp/vars": [ |
| "model_00039.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_40/mlp/shared_experts/down/vars": [ |
| "model_00039.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_40/mlp/shared_experts/gate/vars": [ |
| "model_00039.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_40/mlp/shared_experts/up/vars": [ |
| "model_00039.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_40/post_attention_layernorm/vars": [ |
| "model_00039.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_40/attention/key/vars": [ |
| "model_00039.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_40/attention/output_proj/vars": [ |
| "model_00039.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_40/attention/query/vars": [ |
| "model_00039.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_40/attention/value/vars": [ |
| "model_00039.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_41/input_layernorm/vars": [ |
| "model_00039.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_41/mlp/experts/vars": [ |
| "model_00040.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_41/mlp/vars": [ |
| "model_00040.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_41/mlp/shared_experts/down/vars": [ |
| "model_00040.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_41/mlp/shared_experts/gate/vars": [ |
| "model_00040.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_41/mlp/shared_experts/up/vars": [ |
| "model_00040.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_41/post_attention_layernorm/vars": [ |
| "model_00040.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_41/attention/key/vars": [ |
| "model_00040.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_41/attention/output_proj/vars": [ |
| "model_00040.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_41/attention/query/vars": [ |
| "model_00040.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_41/attention/value/vars": [ |
| "model_00040.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_42/input_layernorm/vars": [ |
| "model_00040.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_42/mlp/experts/vars": [ |
| "model_00041.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_42/mlp/vars": [ |
| "model_00041.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_42/mlp/shared_experts/down/vars": [ |
| "model_00041.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_42/mlp/shared_experts/gate/vars": [ |
| "model_00041.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_42/mlp/shared_experts/up/vars": [ |
| "model_00041.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_42/post_attention_layernorm/vars": [ |
| "model_00041.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_42/attention/key/vars": [ |
| "model_00041.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_42/attention/output_proj/vars": [ |
| "model_00041.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_42/attention/query/vars": [ |
| "model_00041.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_42/attention/value/vars": [ |
| "model_00041.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_43/input_layernorm/vars": [ |
| "model_00041.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_43/mlp/experts/vars": [ |
| "model_00042.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_43/mlp/vars": [ |
| "model_00042.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_43/mlp/shared_experts/down/vars": [ |
| "model_00042.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_43/mlp/shared_experts/gate/vars": [ |
| "model_00042.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_43/mlp/shared_experts/up/vars": [ |
| "model_00042.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_43/post_attention_layernorm/vars": [ |
| "model_00042.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_43/attention/key/vars": [ |
| "model_00042.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_43/attention/output_proj/vars": [ |
| "model_00042.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_43/attention/query/vars": [ |
| "model_00042.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_43/attention/value/vars": [ |
| "model_00042.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_44/input_layernorm/vars": [ |
| "model_00042.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_44/mlp/experts/vars": [ |
| "model_00043.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_44/mlp/vars": [ |
| "model_00043.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_44/mlp/shared_experts/down/vars": [ |
| "model_00043.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_44/mlp/shared_experts/gate/vars": [ |
| "model_00043.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_44/mlp/shared_experts/up/vars": [ |
| "model_00043.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_44/post_attention_layernorm/vars": [ |
| "model_00043.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_44/attention/key/vars": [ |
| "model_00043.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_44/attention/output_proj/vars": [ |
| "model_00043.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_44/attention/query/vars": [ |
| "model_00043.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_44/attention/value/vars": [ |
| "model_00043.weights.h5" |
| ], |
| "/layers/dense/vars": [ |
| "model_00044.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_45/input_layernorm/vars": [ |
| "model_00044.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_45/mlp/experts/vars": [ |
| "model_00044.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_45/mlp/vars": [ |
| "model_00045.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_45/mlp/shared_experts/down/vars": [ |
| "model_00045.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_45/mlp/shared_experts/gate/vars": [ |
| "model_00045.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_45/mlp/shared_experts/up/vars": [ |
| "model_00045.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_45/post_attention_layernorm/vars": [ |
| "model_00045.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_45/attention/key/vars": [ |
| "model_00045.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_45/attention/output_proj/vars": [ |
| "model_00045.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_45/attention/query/vars": [ |
| "model_00045.weights.h5" |
| ], |
| "/layers/glm4_moe_decoder_layer_45/attention/value/vars": [ |
| "model_00045.weights.h5" |
| ], |
| "/layers/glm4_moe_rms_norm/vars": [ |
| "model_00045.weights.h5" |
| ] |
| } |
| } |
