Hugging Face's logo

keras
/
qwen2.5_instruct_32b_en

Text Generation
KerasHub
Model card Files Community
Divyasreepat commited on
Commit
6210e82
·
verified ·
1 Parent(s): db132fe

Upload folder using huggingface_hub

Browse files
Files changed (22) hide show
  1. README.md +25 -0
  2. assets/tokenizer/merges.txt +0 -0
  3. assets/tokenizer/vocabulary.json +0 -0
  4. config.json +22 -0
  5. metadata.json +9 -0
  6. model.weights.json +585 -0
  7. model_00000.weights.h5 +3 -0
  8. model_00001.weights.h5 +3 -0
  9. model_00002.weights.h5 +3 -0
  10. model_00003.weights.h5 +3 -0
  11. model_00004.weights.h5 +3 -0
  12. model_00005.weights.h5 +3 -0
  13. model_00006.weights.h5 +3 -0
  14. model_00007.weights.h5 +3 -0
  15. model_00008.weights.h5 +3 -0
  16. model_00009.weights.h5 +3 -0
  17. model_00010.weights.h5 +3 -0
  18. model_00011.weights.h5 +3 -0
  19. model_00012.weights.h5 +3 -0
  20. preprocessor.json +65 -0
  21. task.json +95 -0
  22. tokenizer.json +44 -0
README.md ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: keras-hub
3
+ pipeline_tag: text-generation
4
+ ---
5
+ This is a [`Qwen` model](https://keras.io/api/keras_hub/models/qwen) uploaded using the KerasHub library and can be used with JAX, TensorFlow, and PyTorch backends.
6
+ This model is related to a `CausalLM` task.
7
+
8
+ Model config:
9
+ * **name:** qwen_backbone
10
+ * **trainable:** True
11
+ * **vocabulary_size:** 152064
12
+ * **num_layers:** 64
13
+ * **num_query_heads:** 40
14
+ * **hidden_dim:** 5120
15
+ * **intermediate_dim:** 27648
16
+ * **rope_max_wavelength:** 1000000.0
17
+ * **rope_scaling_factor:** 1.0
18
+ * **num_key_value_heads:** 8
19
+ * **layer_norm_epsilon:** 1e-06
20
+ * **dropout:** 0
21
+ * **tie_word_embeddings:** False
22
+ * **use_sliding_window_attention:** False
23
+ * **sliding_window_size:** 131072
24
+
25
+ This model card has been generated automatically and should be completed by the model author. See [Model Cards documentation](https://huggingface.co/docs/hub/model-cards) for more information.
assets/tokenizer/merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
assets/tokenizer/vocabulary.json ADDED
The diff for this file is too large to render. See raw diff
 
config.json ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "module": "keras_hub.src.models.qwen.qwen_backbone",
3
+ "class_name": "QwenBackbone",
4
+ "config": {
5
+ "name": "qwen_backbone",
6
+ "trainable": true,
7
+ "vocabulary_size": 152064,
8
+ "num_layers": 64,
9
+ "num_query_heads": 40,
10
+ "hidden_dim": 5120,
11
+ "intermediate_dim": 27648,
12
+ "rope_max_wavelength": 1000000.0,
13
+ "rope_scaling_factor": 1.0,
14
+ "num_key_value_heads": 8,
15
+ "layer_norm_epsilon": 1e-06,
16
+ "dropout": 0,
17
+ "tie_word_embeddings": false,
18
+ "use_sliding_window_attention": false,
19
+ "sliding_window_size": 131072
20
+ },
21
+ "registered_name": "keras_hub>QwenBackbone"
22
+ }
metadata.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "keras_version": "3.10.0.dev2025051003",
3
+ "keras_hub_version": "0.21.0.dev0",
4
+ "parameter_count": 32763876352,
5
+ "date_saved": "2025-05-11@02:05:52",
6
+ "tasks": [
7
+ "CausalLM"
8
+ ]
9
+ }
model.weights.json ADDED
@@ -0,0 +1,585 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 131055505408.0
4
+ },
5
+ "weight_map": {
6
+ "/layers/reversible_embedding/vars": "model_00000.weights.h5",
7
+ "/layers/qwen_transformer_decoder/_feedforward_gate_dense/vars": "model_00000.weights.h5",
8
+ "/layers/qwen_transformer_decoder/_feedforward_intermediate_dense/vars": "model_00000.weights.h5",
9
+ "/layers/qwen_transformer_decoder/_feedforward_layernorm/vars": "model_00000.weights.h5",
10
+ "/layers/qwen_transformer_decoder/_feedforward_output_dense/vars": "model_00000.weights.h5",
11
+ "/layers/qwen_transformer_decoder/_self_attention_layer/_key_dense/vars": "model_00000.weights.h5",
12
+ "/layers/qwen_transformer_decoder/_self_attention_layer/_output_dense/vars": "model_00000.weights.h5",
13
+ "/layers/qwen_transformer_decoder/_self_attention_layer/_query_dense/vars": "model_00000.weights.h5",
14
+ "/layers/qwen_transformer_decoder/_self_attention_layer/_value_dense/vars": "model_00000.weights.h5",
15
+ "/layers/qwen_transformer_decoder/_self_attention_layernorm/vars": "model_00000.weights.h5",
16
+ "/layers/qwen_transformer_decoder_1/_feedforward_gate_dense/vars": "model_00000.weights.h5",
17
+ "/layers/qwen_transformer_decoder_1/_feedforward_intermediate_dense/vars": "model_00000.weights.h5",
18
+ "/layers/qwen_transformer_decoder_1/_feedforward_layernorm/vars": "model_00000.weights.h5",
19
+ "/layers/qwen_transformer_decoder_1/_feedforward_output_dense/vars": "model_00000.weights.h5",
20
+ "/layers/qwen_transformer_decoder_1/_self_attention_layer/_key_dense/vars": "model_00000.weights.h5",
21
+ "/layers/qwen_transformer_decoder_1/_self_attention_layer/_output_dense/vars": "model_00000.weights.h5",
22
+ "/layers/qwen_transformer_decoder_1/_self_attention_layer/_query_dense/vars": "model_00000.weights.h5",
23
+ "/layers/qwen_transformer_decoder_1/_self_attention_layer/_value_dense/vars": "model_00000.weights.h5",
24
+ "/layers/qwen_transformer_decoder_1/_self_attention_layernorm/vars": "model_00000.weights.h5",
25
+ "/layers/qwen_transformer_decoder_2/_feedforward_gate_dense/vars": "model_00000.weights.h5",
26
+ "/layers/qwen_transformer_decoder_2/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
27
+ "/layers/qwen_transformer_decoder_2/_feedforward_layernorm/vars": "model_00001.weights.h5",
28
+ "/layers/qwen_transformer_decoder_2/_feedforward_output_dense/vars": "model_00001.weights.h5",
29
+ "/layers/qwen_transformer_decoder_2/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
30
+ "/layers/qwen_transformer_decoder_2/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
31
+ "/layers/qwen_transformer_decoder_2/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
32
+ "/layers/qwen_transformer_decoder_2/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
33
+ "/layers/qwen_transformer_decoder_2/_self_attention_layernorm/vars": "model_00001.weights.h5",
34
+ "/layers/qwen_transformer_decoder_3/_feedforward_gate_dense/vars": "model_00001.weights.h5",
35
+ "/layers/qwen_transformer_decoder_3/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
36
+ "/layers/qwen_transformer_decoder_3/_feedforward_layernorm/vars": "model_00001.weights.h5",
37
+ "/layers/qwen_transformer_decoder_3/_feedforward_output_dense/vars": "model_00001.weights.h5",
38
+ "/layers/qwen_transformer_decoder_3/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
39
+ "/layers/qwen_transformer_decoder_3/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
40
+ "/layers/qwen_transformer_decoder_3/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
41
+ "/layers/qwen_transformer_decoder_3/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
42
+ "/layers/qwen_transformer_decoder_3/_self_attention_layernorm/vars": "model_00001.weights.h5",
43
+ "/layers/qwen_transformer_decoder_4/_feedforward_gate_dense/vars": "model_00001.weights.h5",
44
+ "/layers/qwen_transformer_decoder_4/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
45
+ "/layers/qwen_transformer_decoder_4/_feedforward_layernorm/vars": "model_00001.weights.h5",
46
+ "/layers/qwen_transformer_decoder_4/_feedforward_output_dense/vars": "model_00001.weights.h5",
47
+ "/layers/qwen_transformer_decoder_4/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
48
+ "/layers/qwen_transformer_decoder_4/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
49
+ "/layers/qwen_transformer_decoder_4/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
50
+ "/layers/qwen_transformer_decoder_4/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
51
+ "/layers/qwen_transformer_decoder_4/_self_attention_layernorm/vars": "model_00001.weights.h5",
52
+ "/layers/qwen_transformer_decoder_5/_feedforward_gate_dense/vars": "model_00001.weights.h5",
53
+ "/layers/qwen_transformer_decoder_5/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
54
+ "/layers/qwen_transformer_decoder_5/_feedforward_layernorm/vars": "model_00001.weights.h5",
55
+ "/layers/qwen_transformer_decoder_5/_feedforward_output_dense/vars": "model_00001.weights.h5",
56
+ "/layers/qwen_transformer_decoder_5/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
57
+ "/layers/qwen_transformer_decoder_5/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
58
+ "/layers/qwen_transformer_decoder_5/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
59
+ "/layers/qwen_transformer_decoder_5/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
60
+ "/layers/qwen_transformer_decoder_5/_self_attention_layernorm/vars": "model_00001.weights.h5",
61
+ "/layers/qwen_transformer_decoder_6/_feedforward_gate_dense/vars": "model_00001.weights.h5",
62
+ "/layers/qwen_transformer_decoder_6/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
63
+ "/layers/qwen_transformer_decoder_6/_feedforward_layernorm/vars": "model_00001.weights.h5",
64
+ "/layers/qwen_transformer_decoder_6/_feedforward_output_dense/vars": "model_00001.weights.h5",
65
+ "/layers/qwen_transformer_decoder_6/_self_attention_layer/_key_dense/vars": "model_00001.weights.h5",
66
+ "/layers/qwen_transformer_decoder_6/_self_attention_layer/_output_dense/vars": "model_00001.weights.h5",
67
+ "/layers/qwen_transformer_decoder_6/_self_attention_layer/_query_dense/vars": "model_00001.weights.h5",
68
+ "/layers/qwen_transformer_decoder_6/_self_attention_layer/_value_dense/vars": "model_00001.weights.h5",
69
+ "/layers/qwen_transformer_decoder_6/_self_attention_layernorm/vars": "model_00001.weights.h5",
70
+ "/layers/qwen_transformer_decoder_7/_feedforward_gate_dense/vars": "model_00001.weights.h5",
71
+ "/layers/qwen_transformer_decoder_7/_feedforward_intermediate_dense/vars": "model_00001.weights.h5",
72
+ "/layers/qwen_transformer_decoder_7/_feedforward_layernorm/vars": "model_00001.weights.h5",
73
+ "/layers/qwen_transformer_decoder_7/_feedforward_output_dense/vars": "model_00002.weights.h5",
74
+ "/layers/qwen_transformer_decoder_7/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
75
+ "/layers/qwen_transformer_decoder_7/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
76
+ "/layers/qwen_transformer_decoder_7/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
77
+ "/layers/qwen_transformer_decoder_7/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
78
+ "/layers/qwen_transformer_decoder_7/_self_attention_layernorm/vars": "model_00002.weights.h5",
79
+ "/layers/qwen_transformer_decoder_8/_feedforward_gate_dense/vars": "model_00002.weights.h5",
80
+ "/layers/qwen_transformer_decoder_8/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
81
+ "/layers/qwen_transformer_decoder_8/_feedforward_layernorm/vars": "model_00002.weights.h5",
82
+ "/layers/qwen_transformer_decoder_8/_feedforward_output_dense/vars": "model_00002.weights.h5",
83
+ "/layers/qwen_transformer_decoder_8/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
84
+ "/layers/qwen_transformer_decoder_8/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
85
+ "/layers/qwen_transformer_decoder_8/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
86
+ "/layers/qwen_transformer_decoder_8/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
87
+ "/layers/qwen_transformer_decoder_8/_self_attention_layernorm/vars": "model_00002.weights.h5",
88
+ "/layers/qwen_transformer_decoder_9/_feedforward_gate_dense/vars": "model_00002.weights.h5",
89
+ "/layers/qwen_transformer_decoder_9/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
90
+ "/layers/qwen_transformer_decoder_9/_feedforward_layernorm/vars": "model_00002.weights.h5",
91
+ "/layers/qwen_transformer_decoder_9/_feedforward_output_dense/vars": "model_00002.weights.h5",
92
+ "/layers/qwen_transformer_decoder_9/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
93
+ "/layers/qwen_transformer_decoder_9/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
94
+ "/layers/qwen_transformer_decoder_9/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
95
+ "/layers/qwen_transformer_decoder_9/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
96
+ "/layers/qwen_transformer_decoder_9/_self_attention_layernorm/vars": "model_00002.weights.h5",
97
+ "/layers/qwen_transformer_decoder_10/_feedforward_gate_dense/vars": "model_00002.weights.h5",
98
+ "/layers/qwen_transformer_decoder_10/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
99
+ "/layers/qwen_transformer_decoder_10/_feedforward_layernorm/vars": "model_00002.weights.h5",
100
+ "/layers/qwen_transformer_decoder_10/_feedforward_output_dense/vars": "model_00002.weights.h5",
101
+ "/layers/qwen_transformer_decoder_10/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
102
+ "/layers/qwen_transformer_decoder_10/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
103
+ "/layers/qwen_transformer_decoder_10/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
104
+ "/layers/qwen_transformer_decoder_10/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
105
+ "/layers/qwen_transformer_decoder_10/_self_attention_layernorm/vars": "model_00002.weights.h5",
106
+ "/layers/qwen_transformer_decoder_11/_feedforward_gate_dense/vars": "model_00002.weights.h5",
107
+ "/layers/qwen_transformer_decoder_11/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
108
+ "/layers/qwen_transformer_decoder_11/_feedforward_layernorm/vars": "model_00002.weights.h5",
109
+ "/layers/qwen_transformer_decoder_11/_feedforward_output_dense/vars": "model_00002.weights.h5",
110
+ "/layers/qwen_transformer_decoder_11/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
111
+ "/layers/qwen_transformer_decoder_11/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
112
+ "/layers/qwen_transformer_decoder_11/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
113
+ "/layers/qwen_transformer_decoder_11/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
114
+ "/layers/qwen_transformer_decoder_11/_self_attention_layernorm/vars": "model_00002.weights.h5",
115
+ "/layers/qwen_transformer_decoder_12/_feedforward_gate_dense/vars": "model_00002.weights.h5",
116
+ "/layers/qwen_transformer_decoder_12/_feedforward_intermediate_dense/vars": "model_00002.weights.h5",
117
+ "/layers/qwen_transformer_decoder_12/_feedforward_layernorm/vars": "model_00002.weights.h5",
118
+ "/layers/qwen_transformer_decoder_12/_feedforward_output_dense/vars": "model_00002.weights.h5",
119
+ "/layers/qwen_transformer_decoder_12/_self_attention_layer/_key_dense/vars": "model_00002.weights.h5",
120
+ "/layers/qwen_transformer_decoder_12/_self_attention_layer/_output_dense/vars": "model_00002.weights.h5",
121
+ "/layers/qwen_transformer_decoder_12/_self_attention_layer/_query_dense/vars": "model_00002.weights.h5",
122
+ "/layers/qwen_transformer_decoder_12/_self_attention_layer/_value_dense/vars": "model_00002.weights.h5",
123
+ "/layers/qwen_transformer_decoder_12/_self_attention_layernorm/vars": "model_00002.weights.h5",
124
+ "/layers/qwen_transformer_decoder_13/_feedforward_gate_dense/vars": "model_00003.weights.h5",
125
+ "/layers/qwen_transformer_decoder_13/_feedforward_intermediate_dense/vars": "model_00003.weights.h5",
126
+ "/layers/qwen_transformer_decoder_13/_feedforward_layernorm/vars": "model_00003.weights.h5",
127
+ "/layers/qwen_transformer_decoder_13/_feedforward_output_dense/vars": "model_00003.weights.h5",
128
+ "/layers/qwen_transformer_decoder_13/_self_attention_layer/_key_dense/vars": "model_00003.weights.h5",
129
+ "/layers/qwen_transformer_decoder_13/_self_attention_layer/_output_dense/vars": "model_00003.weights.h5",
130
+ "/layers/qwen_transformer_decoder_13/_self_attention_layer/_query_dense/vars": "model_00003.weights.h5",
131
+ "/layers/qwen_transformer_decoder_13/_self_attention_layer/_value_dense/vars": "model_00003.weights.h5",
132
+ "/layers/qwen_transformer_decoder_13/_self_attention_layernorm/vars": "model_00003.weights.h5",
133
+ "/layers/qwen_transformer_decoder_14/_feedforward_gate_dense/vars": "model_00003.weights.h5",
134
+ "/layers/qwen_transformer_decoder_14/_feedforward_intermediate_dense/vars": "model_00003.weights.h5",
135
+ "/layers/qwen_transformer_decoder_14/_feedforward_layernorm/vars": "model_00003.weights.h5",
136
+ "/layers/qwen_transformer_decoder_14/_feedforward_output_dense/vars": "model_00003.weights.h5",
137
+ "/layers/qwen_transformer_decoder_14/_self_attention_layer/_key_dense/vars": "model_00003.weights.h5",
138
+ "/layers/qwen_transformer_decoder_14/_self_attention_layer/_output_dense/vars": "model_00003.weights.h5",
139
+ "/layers/qwen_transformer_decoder_14/_self_attention_layer/_query_dense/vars": "model_00003.weights.h5",
140
+ "/layers/qwen_transformer_decoder_14/_self_attention_layer/_value_dense/vars": "model_00003.weights.h5",
141
+ "/layers/qwen_transformer_decoder_14/_self_attention_layernorm/vars": "model_00003.weights.h5",
142
+ "/layers/qwen_transformer_decoder_15/_feedforward_gate_dense/vars": "model_00003.weights.h5",
143
+ "/layers/qwen_transformer_decoder_15/_feedforward_intermediate_dense/vars": "model_00003.weights.h5",
144
+ "/layers/qwen_transformer_decoder_15/_feedforward_layernorm/vars": "model_00003.weights.h5",
145
+ "/layers/qwen_transformer_decoder_15/_feedforward_output_dense/vars": "model_00003.weights.h5",
146
+ "/layers/qwen_transformer_decoder_15/_self_attention_layer/_key_dense/vars": "model_00003.weights.h5",
147
+ "/layers/qwen_transformer_decoder_15/_self_attention_layer/_output_dense/vars": "model_00003.weights.h5",
148
+ "/layers/qwen_transformer_decoder_15/_self_attention_layer/_query_dense/vars": "model_00003.weights.h5",
149
+ "/layers/qwen_transformer_decoder_15/_self_attention_layer/_value_dense/vars": "model_00003.weights.h5",
150
+ "/layers/qwen_transformer_decoder_15/_self_attention_layernorm/vars": "model_00003.weights.h5",
151
+ "/layers/qwen_transformer_decoder_16/_feedforward_gate_dense/vars": "model_00003.weights.h5",
152
+ "/layers/qwen_transformer_decoder_16/_feedforward_intermediate_dense/vars": "model_00003.weights.h5",
153
+ "/layers/qwen_transformer_decoder_16/_feedforward_layernorm/vars": "model_00003.weights.h5",
154
+ "/layers/qwen_transformer_decoder_16/_feedforward_output_dense/vars": "model_00003.weights.h5",
155
+ "/layers/qwen_transformer_decoder_16/_self_attention_layer/_key_dense/vars": "model_00003.weights.h5",
156
+ "/layers/qwen_transformer_decoder_16/_self_attention_layer/_output_dense/vars": "model_00003.weights.h5",
157
+ "/layers/qwen_transformer_decoder_16/_self_attention_layer/_query_dense/vars": "model_00003.weights.h5",
158
+ "/layers/qwen_transformer_decoder_16/_self_attention_layer/_value_dense/vars": "model_00003.weights.h5",
159
+ "/layers/qwen_transformer_decoder_16/_self_attention_layernorm/vars": "model_00003.weights.h5",
160
+ "/layers/qwen_transformer_decoder_17/_feedforward_gate_dense/vars": "model_00003.weights.h5",
161
+ "/layers/qwen_transformer_decoder_17/_feedforward_intermediate_dense/vars": "model_00003.weights.h5",
162
+ "/layers/qwen_transformer_decoder_17/_feedforward_layernorm/vars": "model_00003.weights.h5",
163
+ "/layers/qwen_transformer_decoder_17/_feedforward_output_dense/vars": "model_00003.weights.h5",
164
+ "/layers/qwen_transformer_decoder_17/_self_attention_layer/_key_dense/vars": "model_00003.weights.h5",
165
+ "/layers/qwen_transformer_decoder_17/_self_attention_layer/_output_dense/vars": "model_00003.weights.h5",
166
+ "/layers/qwen_transformer_decoder_17/_self_attention_layer/_query_dense/vars": "model_00003.weights.h5",
167
+ "/layers/qwen_transformer_decoder_17/_self_attention_layer/_value_dense/vars": "model_00003.weights.h5",
168
+ "/layers/qwen_transformer_decoder_17/_self_attention_layernorm/vars": "model_00003.weights.h5",
169
+ "/layers/qwen_transformer_decoder_18/_feedforward_gate_dense/vars": "model_00003.weights.h5",
170
+ "/layers/qwen_transformer_decoder_18/_feedforward_intermediate_dense/vars": "model_00004.weights.h5",
171
+ "/layers/qwen_transformer_decoder_18/_feedforward_layernorm/vars": "model_00004.weights.h5",
172
+ "/layers/qwen_transformer_decoder_18/_feedforward_output_dense/vars": "model_00004.weights.h5",
173
+ "/layers/qwen_transformer_decoder_18/_self_attention_layer/_key_dense/vars": "model_00004.weights.h5",
174
+ "/layers/qwen_transformer_decoder_18/_self_attention_layer/_output_dense/vars": "model_00004.weights.h5",
175
+ "/layers/qwen_transformer_decoder_18/_self_attention_layer/_query_dense/vars": "model_00004.weights.h5",
176
+ "/layers/qwen_transformer_decoder_18/_self_attention_layer/_value_dense/vars": "model_00004.weights.h5",
177
+ "/layers/qwen_transformer_decoder_18/_self_attention_layernorm/vars": "model_00004.weights.h5",
178
+ "/layers/qwen_transformer_decoder_19/_feedforward_gate_dense/vars": "model_00004.weights.h5",
179
+ "/layers/qwen_transformer_decoder_19/_feedforward_intermediate_dense/vars": "model_00004.weights.h5",
180
+ "/layers/qwen_transformer_decoder_19/_feedforward_layernorm/vars": "model_00004.weights.h5",
181
+ "/layers/qwen_transformer_decoder_19/_feedforward_output_dense/vars": "model_00004.weights.h5",
182
+ "/layers/qwen_transformer_decoder_19/_self_attention_layer/_key_dense/vars": "model_00004.weights.h5",
183
+ "/layers/qwen_transformer_decoder_19/_self_attention_layer/_output_dense/vars": "model_00004.weights.h5",
184
+ "/layers/qwen_transformer_decoder_19/_self_attention_layer/_query_dense/vars": "model_00004.weights.h5",
185
+ "/layers/qwen_transformer_decoder_19/_self_attention_layer/_value_dense/vars": "model_00004.weights.h5",
186
+ "/layers/qwen_transformer_decoder_19/_self_attention_layernorm/vars": "model_00004.weights.h5",
187
+ "/layers/qwen_transformer_decoder_20/_feedforward_gate_dense/vars": "model_00004.weights.h5",
188
+ "/layers/qwen_transformer_decoder_20/_feedforward_intermediate_dense/vars": "model_00004.weights.h5",
189
+ "/layers/qwen_transformer_decoder_20/_feedforward_layernorm/vars": "model_00004.weights.h5",
190
+ "/layers/qwen_transformer_decoder_20/_feedforward_output_dense/vars": "model_00004.weights.h5",
191
+ "/layers/qwen_transformer_decoder_20/_self_attention_layer/_key_dense/vars": "model_00004.weights.h5",
192
+ "/layers/qwen_transformer_decoder_20/_self_attention_layer/_output_dense/vars": "model_00004.weights.h5",
193
+ "/layers/qwen_transformer_decoder_20/_self_attention_layer/_query_dense/vars": "model_00004.weights.h5",
194
+ "/layers/qwen_transformer_decoder_20/_self_attention_layer/_value_dense/vars": "model_00004.weights.h5",
195
+ "/layers/qwen_transformer_decoder_20/_self_attention_layernorm/vars": "model_00004.weights.h5",
196
+ "/layers/qwen_transformer_decoder_21/_feedforward_gate_dense/vars": "model_00004.weights.h5",
197
+ "/layers/qwen_transformer_decoder_21/_feedforward_intermediate_dense/vars": "model_00004.weights.h5",
198
+ "/layers/qwen_transformer_decoder_21/_feedforward_layernorm/vars": "model_00004.weights.h5",
199
+ "/layers/qwen_transformer_decoder_21/_feedforward_output_dense/vars": "model_00004.weights.h5",
200
+ "/layers/qwen_transformer_decoder_21/_self_attention_layer/_key_dense/vars": "model_00004.weights.h5",
201
+ "/layers/qwen_transformer_decoder_21/_self_attention_layer/_output_dense/vars": "model_00004.weights.h5",
202
+ "/layers/qwen_transformer_decoder_21/_self_attention_layer/_query_dense/vars": "model_00004.weights.h5",
203
+ "/layers/qwen_transformer_decoder_21/_self_attention_layer/_value_dense/vars": "model_00004.weights.h5",
204
+ "/layers/qwen_transformer_decoder_21/_self_attention_layernorm/vars": "model_00004.weights.h5",
205
+ "/layers/qwen_transformer_decoder_22/_feedforward_gate_dense/vars": "model_00004.weights.h5",
206
+ "/layers/qwen_transformer_decoder_22/_feedforward_intermediate_dense/vars": "model_00004.weights.h5",
207
+ "/layers/qwen_transformer_decoder_22/_feedforward_layernorm/vars": "model_00004.weights.h5",
208
+ "/layers/qwen_transformer_decoder_22/_feedforward_output_dense/vars": "model_00004.weights.h5",
209
+ "/layers/qwen_transformer_decoder_22/_self_attention_layer/_key_dense/vars": "model_00004.weights.h5",
210
+ "/layers/qwen_transformer_decoder_22/_self_attention_layer/_output_dense/vars": "model_00004.weights.h5",
211
+ "/layers/qwen_transformer_decoder_22/_self_attention_layer/_query_dense/vars": "model_00004.weights.h5",
212
+ "/layers/qwen_transformer_decoder_22/_self_attention_layer/_value_dense/vars": "model_00004.weights.h5",
213
+ "/layers/qwen_transformer_decoder_22/_self_attention_layernorm/vars": "model_00004.weights.h5",
214
+ "/layers/qwen_transformer_decoder_23/_feedforward_gate_dense/vars": "model_00004.weights.h5",
215
+ "/layers/qwen_transformer_decoder_23/_feedforward_intermediate_dense/vars": "model_00004.weights.h5",
216
+ "/layers/qwen_transformer_decoder_23/_feedforward_layernorm/vars": "model_00004.weights.h5",
217
+ "/layers/qwen_transformer_decoder_23/_feedforward_output_dense/vars": "model_00005.weights.h5",
218
+ "/layers/qwen_transformer_decoder_23/_self_attention_layer/_key_dense/vars": "model_00005.weights.h5",
219
+ "/layers/qwen_transformer_decoder_23/_self_attention_layer/_output_dense/vars": "model_00005.weights.h5",
220
+ "/layers/qwen_transformer_decoder_23/_self_attention_layer/_query_dense/vars": "model_00005.weights.h5",
221
+ "/layers/qwen_transformer_decoder_23/_self_attention_layer/_value_dense/vars": "model_00005.weights.h5",
222
+ "/layers/qwen_transformer_decoder_23/_self_attention_layernorm/vars": "model_00005.weights.h5",
223
+ "/layers/qwen_transformer_decoder_24/_feedforward_gate_dense/vars": "model_00005.weights.h5",
224
+ "/layers/qwen_transformer_decoder_24/_feedforward_intermediate_dense/vars": "model_00005.weights.h5",
225
+ "/layers/qwen_transformer_decoder_24/_feedforward_layernorm/vars": "model_00005.weights.h5",
226
+ "/layers/qwen_transformer_decoder_24/_feedforward_output_dense/vars": "model_00005.weights.h5",
227
+ "/layers/qwen_transformer_decoder_24/_self_attention_layer/_key_dense/vars": "model_00005.weights.h5",
228
+ "/layers/qwen_transformer_decoder_24/_self_attention_layer/_output_dense/vars": "model_00005.weights.h5",
229
+ "/layers/qwen_transformer_decoder_24/_self_attention_layer/_query_dense/vars": "model_00005.weights.h5",
230
+ "/layers/qwen_transformer_decoder_24/_self_attention_layer/_value_dense/vars": "model_00005.weights.h5",
231
+ "/layers/qwen_transformer_decoder_24/_self_attention_layernorm/vars": "model_00005.weights.h5",
232
+ "/layers/qwen_transformer_decoder_25/_feedforward_gate_dense/vars": "model_00005.weights.h5",
233
+ "/layers/qwen_transformer_decoder_25/_feedforward_intermediate_dense/vars": "model_00005.weights.h5",
234
+ "/layers/qwen_transformer_decoder_25/_feedforward_layernorm/vars": "model_00005.weights.h5",
235
+ "/layers/qwen_transformer_decoder_25/_feedforward_output_dense/vars": "model_00005.weights.h5",
236
+ "/layers/qwen_transformer_decoder_25/_self_attention_layer/_key_dense/vars": "model_00005.weights.h5",
237
+ "/layers/qwen_transformer_decoder_25/_self_attention_layer/_output_dense/vars": "model_00005.weights.h5",
238
+ "/layers/qwen_transformer_decoder_25/_self_attention_layer/_query_dense/vars": "model_00005.weights.h5",
239
+ "/layers/qwen_transformer_decoder_25/_self_attention_layer/_value_dense/vars": "model_00005.weights.h5",
240
+ "/layers/qwen_transformer_decoder_25/_self_attention_layernorm/vars": "model_00005.weights.h5",
241
+ "/layers/qwen_transformer_decoder_26/_feedforward_gate_dense/vars": "model_00005.weights.h5",
242
+ "/layers/qwen_transformer_decoder_26/_feedforward_intermediate_dense/vars": "model_00005.weights.h5",
243
+ "/layers/qwen_transformer_decoder_26/_feedforward_layernorm/vars": "model_00005.weights.h5",
244
+ "/layers/qwen_transformer_decoder_26/_feedforward_output_dense/vars": "model_00005.weights.h5",
245
+ "/layers/qwen_transformer_decoder_26/_self_attention_layer/_key_dense/vars": "model_00005.weights.h5",
246
+ "/layers/qwen_transformer_decoder_26/_self_attention_layer/_output_dense/vars": "model_00005.weights.h5",
247
+ "/layers/qwen_transformer_decoder_26/_self_attention_layer/_query_dense/vars": "model_00005.weights.h5",
248
+ "/layers/qwen_transformer_decoder_26/_self_attention_layer/_value_dense/vars": "model_00005.weights.h5",
249
+ "/layers/qwen_transformer_decoder_26/_self_attention_layernorm/vars": "model_00005.weights.h5",
250
+ "/layers/qwen_transformer_decoder_27/_feedforward_gate_dense/vars": "model_00005.weights.h5",
251
+ "/layers/qwen_transformer_decoder_27/_feedforward_intermediate_dense/vars": "model_00005.weights.h5",
252
+ "/layers/qwen_transformer_decoder_27/_feedforward_layernorm/vars": "model_00005.weights.h5",
253
+ "/layers/qwen_transformer_decoder_27/_feedforward_output_dense/vars": "model_00005.weights.h5",
254
+ "/layers/qwen_transformer_decoder_27/_self_attention_layer/_key_dense/vars": "model_00005.weights.h5",
255
+ "/layers/qwen_transformer_decoder_27/_self_attention_layer/_output_dense/vars": "model_00005.weights.h5",
256
+ "/layers/qwen_transformer_decoder_27/_self_attention_layer/_query_dense/vars": "model_00005.weights.h5",
257
+ "/layers/qwen_transformer_decoder_27/_self_attention_layer/_value_dense/vars": "model_00005.weights.h5",
258
+ "/layers/qwen_transformer_decoder_27/_self_attention_layernorm/vars": "model_00005.weights.h5",
259
+ "/layers/qwen_transformer_decoder_28/_feedforward_gate_dense/vars": "model_00005.weights.h5",
260
+ "/layers/qwen_transformer_decoder_28/_feedforward_intermediate_dense/vars": "model_00005.weights.h5",
261
+ "/layers/qwen_transformer_decoder_28/_feedforward_layernorm/vars": "model_00005.weights.h5",
262
+ "/layers/qwen_transformer_decoder_28/_feedforward_output_dense/vars": "model_00005.weights.h5",
263
+ "/layers/qwen_transformer_decoder_28/_self_attention_layer/_key_dense/vars": "model_00005.weights.h5",
264
+ "/layers/qwen_transformer_decoder_28/_self_attention_layer/_output_dense/vars": "model_00005.weights.h5",
265
+ "/layers/qwen_transformer_decoder_28/_self_attention_layer/_query_dense/vars": "model_00005.weights.h5",
266
+ "/layers/qwen_transformer_decoder_28/_self_attention_layer/_value_dense/vars": "model_00005.weights.h5",
267
+ "/layers/qwen_transformer_decoder_28/_self_attention_layernorm/vars": "model_00005.weights.h5",
268
+ "/layers/qwen_transformer_decoder_29/_feedforward_gate_dense/vars": "model_00006.weights.h5",
269
+ "/layers/qwen_transformer_decoder_29/_feedforward_intermediate_dense/vars": "model_00006.weights.h5",
270
+ "/layers/qwen_transformer_decoder_29/_feedforward_layernorm/vars": "model_00006.weights.h5",
271
+ "/layers/qwen_transformer_decoder_29/_feedforward_output_dense/vars": "model_00006.weights.h5",
272
+ "/layers/qwen_transformer_decoder_29/_self_attention_layer/_key_dense/vars": "model_00006.weights.h5",
273
+ "/layers/qwen_transformer_decoder_29/_self_attention_layer/_output_dense/vars": "model_00006.weights.h5",
274
+ "/layers/qwen_transformer_decoder_29/_self_attention_layer/_query_dense/vars": "model_00006.weights.h5",
275
+ "/layers/qwen_transformer_decoder_29/_self_attention_layer/_value_dense/vars": "model_00006.weights.h5",
276
+ "/layers/qwen_transformer_decoder_29/_self_attention_layernorm/vars": "model_00006.weights.h5",
277
+ "/layers/qwen_transformer_decoder_30/_feedforward_gate_dense/vars": "model_00006.weights.h5",
278
+ "/layers/qwen_transformer_decoder_30/_feedforward_intermediate_dense/vars": "model_00006.weights.h5",
279
+ "/layers/qwen_transformer_decoder_30/_feedforward_layernorm/vars": "model_00006.weights.h5",
280
+ "/layers/qwen_transformer_decoder_30/_feedforward_output_dense/vars": "model_00006.weights.h5",
281
+ "/layers/qwen_transformer_decoder_30/_self_attention_layer/_key_dense/vars": "model_00006.weights.h5",
282
+ "/layers/qwen_transformer_decoder_30/_self_attention_layer/_output_dense/vars": "model_00006.weights.h5",
283
+ "/layers/qwen_transformer_decoder_30/_self_attention_layer/_query_dense/vars": "model_00006.weights.h5",
284
+ "/layers/qwen_transformer_decoder_30/_self_attention_layer/_value_dense/vars": "model_00006.weights.h5",
285
+ "/layers/qwen_transformer_decoder_30/_self_attention_layernorm/vars": "model_00006.weights.h5",
286
+ "/layers/qwen_transformer_decoder_31/_feedforward_gate_dense/vars": "model_00006.weights.h5",
287
+ "/layers/qwen_transformer_decoder_31/_feedforward_intermediate_dense/vars": "model_00006.weights.h5",
288
+ "/layers/qwen_transformer_decoder_31/_feedforward_layernorm/vars": "model_00006.weights.h5",
289
+ "/layers/qwen_transformer_decoder_31/_feedforward_output_dense/vars": "model_00006.weights.h5",
290
+ "/layers/qwen_transformer_decoder_31/_self_attention_layer/_key_dense/vars": "model_00006.weights.h5",
291
+ "/layers/qwen_transformer_decoder_31/_self_attention_layer/_output_dense/vars": "model_00006.weights.h5",
292
+ "/layers/qwen_transformer_decoder_31/_self_attention_layer/_query_dense/vars": "model_00006.weights.h5",
293
+ "/layers/qwen_transformer_decoder_31/_self_attention_layer/_value_dense/vars": "model_00006.weights.h5",
294
+ "/layers/qwen_transformer_decoder_31/_self_attention_layernorm/vars": "model_00006.weights.h5",
295
+ "/layers/qwen_transformer_decoder_32/_feedforward_gate_dense/vars": "model_00006.weights.h5",
296
+ "/layers/qwen_transformer_decoder_32/_feedforward_intermediate_dense/vars": "model_00006.weights.h5",
297
+ "/layers/qwen_transformer_decoder_32/_feedforward_layernorm/vars": "model_00006.weights.h5",
298
+ "/layers/qwen_transformer_decoder_32/_feedforward_output_dense/vars": "model_00006.weights.h5",
299
+ "/layers/qwen_transformer_decoder_32/_self_attention_layer/_key_dense/vars": "model_00006.weights.h5",
300
+ "/layers/qwen_transformer_decoder_32/_self_attention_layer/_output_dense/vars": "model_00006.weights.h5",
301
+ "/layers/qwen_transformer_decoder_32/_self_attention_layer/_query_dense/vars": "model_00006.weights.h5",
302
+ "/layers/qwen_transformer_decoder_32/_self_attention_layer/_value_dense/vars": "model_00006.weights.h5",
303
+ "/layers/qwen_transformer_decoder_32/_self_attention_layernorm/vars": "model_00006.weights.h5",
304
+ "/layers/qwen_transformer_decoder_33/_feedforward_gate_dense/vars": "model_00006.weights.h5",
305
+ "/layers/qwen_transformer_decoder_33/_feedforward_intermediate_dense/vars": "model_00006.weights.h5",
306
+ "/layers/qwen_transformer_decoder_33/_feedforward_layernorm/vars": "model_00006.weights.h5",
307
+ "/layers/qwen_transformer_decoder_33/_feedforward_output_dense/vars": "model_00006.weights.h5",
308
+ "/layers/qwen_transformer_decoder_33/_self_attention_layer/_key_dense/vars": "model_00006.weights.h5",
309
+ "/layers/qwen_transformer_decoder_33/_self_attention_layer/_output_dense/vars": "model_00006.weights.h5",
310
+ "/layers/qwen_transformer_decoder_33/_self_attention_layer/_query_dense/vars": "model_00006.weights.h5",
311
+ "/layers/qwen_transformer_decoder_33/_self_attention_layer/_value_dense/vars": "model_00006.weights.h5",
312
+ "/layers/qwen_transformer_decoder_33/_self_attention_layernorm/vars": "model_00006.weights.h5",
313
+ "/layers/qwen_transformer_decoder_34/_feedforward_gate_dense/vars": "model_00006.weights.h5",
314
+ "/layers/qwen_transformer_decoder_34/_feedforward_intermediate_dense/vars": "model_00007.weights.h5",
315
+ "/layers/qwen_transformer_decoder_34/_feedforward_layernorm/vars": "model_00007.weights.h5",
316
+ "/layers/qwen_transformer_decoder_34/_feedforward_output_dense/vars": "model_00007.weights.h5",
317
+ "/layers/qwen_transformer_decoder_34/_self_attention_layer/_key_dense/vars": "model_00007.weights.h5",
318
+ "/layers/qwen_transformer_decoder_34/_self_attention_layer/_output_dense/vars": "model_00007.weights.h5",
319
+ "/layers/qwen_transformer_decoder_34/_self_attention_layer/_query_dense/vars": "model_00007.weights.h5",
320
+ "/layers/qwen_transformer_decoder_34/_self_attention_layer/_value_dense/vars": "model_00007.weights.h5",
321
+ "/layers/qwen_transformer_decoder_34/_self_attention_layernorm/vars": "model_00007.weights.h5",
322
+ "/layers/qwen_transformer_decoder_35/_feedforward_gate_dense/vars": "model_00007.weights.h5",
323
+ "/layers/qwen_transformer_decoder_35/_feedforward_intermediate_dense/vars": "model_00007.weights.h5",
324
+ "/layers/qwen_transformer_decoder_35/_feedforward_layernorm/vars": "model_00007.weights.h5",
325
+ "/layers/qwen_transformer_decoder_35/_feedforward_output_dense/vars": "model_00007.weights.h5",
326
+ "/layers/qwen_transformer_decoder_35/_self_attention_layer/_key_dense/vars": "model_00007.weights.h5",
327
+ "/layers/qwen_transformer_decoder_35/_self_attention_layer/_output_dense/vars": "model_00007.weights.h5",
328
+ "/layers/qwen_transformer_decoder_35/_self_attention_layer/_query_dense/vars": "model_00007.weights.h5",
329
+ "/layers/qwen_transformer_decoder_35/_self_attention_layer/_value_dense/vars": "model_00007.weights.h5",
330
+ "/layers/qwen_transformer_decoder_35/_self_attention_layernorm/vars": "model_00007.weights.h5",
331
+ "/layers/qwen_transformer_decoder_36/_feedforward_gate_dense/vars": "model_00007.weights.h5",
332
+ "/layers/qwen_transformer_decoder_36/_feedforward_intermediate_dense/vars": "model_00007.weights.h5",
333
+ "/layers/qwen_transformer_decoder_36/_feedforward_layernorm/vars": "model_00007.weights.h5",
334
+ "/layers/qwen_transformer_decoder_36/_feedforward_output_dense/vars": "model_00007.weights.h5",
335
+ "/layers/qwen_transformer_decoder_36/_self_attention_layer/_key_dense/vars": "model_00007.weights.h5",
336
+ "/layers/qwen_transformer_decoder_36/_self_attention_layer/_output_dense/vars": "model_00007.weights.h5",
337
+ "/layers/qwen_transformer_decoder_36/_self_attention_layer/_query_dense/vars": "model_00007.weights.h5",
338
+ "/layers/qwen_transformer_decoder_36/_self_attention_layer/_value_dense/vars": "model_00007.weights.h5",
339
+ "/layers/qwen_transformer_decoder_36/_self_attention_layernorm/vars": "model_00007.weights.h5",
340
+ "/layers/qwen_transformer_decoder_37/_feedforward_gate_dense/vars": "model_00007.weights.h5",
341
+ "/layers/qwen_transformer_decoder_37/_feedforward_intermediate_dense/vars": "model_00007.weights.h5",
342
+ "/layers/qwen_transformer_decoder_37/_feedforward_layernorm/vars": "model_00007.weights.h5",
343
+ "/layers/qwen_transformer_decoder_37/_feedforward_output_dense/vars": "model_00007.weights.h5",
344
+ "/layers/qwen_transformer_decoder_37/_self_attention_layer/_key_dense/vars": "model_00007.weights.h5",
345
+ "/layers/qwen_transformer_decoder_37/_self_attention_layer/_output_dense/vars": "model_00007.weights.h5",
346
+ "/layers/qwen_transformer_decoder_37/_self_attention_layer/_query_dense/vars": "model_00007.weights.h5",
347
+ "/layers/qwen_transformer_decoder_37/_self_attention_layer/_value_dense/vars": "model_00007.weights.h5",
348
+ "/layers/qwen_transformer_decoder_37/_self_attention_layernorm/vars": "model_00007.weights.h5",
349
+ "/layers/qwen_transformer_decoder_38/_feedforward_gate_dense/vars": "model_00007.weights.h5",
350
+ "/layers/qwen_transformer_decoder_38/_feedforward_intermediate_dense/vars": "model_00007.weights.h5",
351
+ "/layers/qwen_transformer_decoder_38/_feedforward_layernorm/vars": "model_00007.weights.h5",
352
+ "/layers/qwen_transformer_decoder_38/_feedforward_output_dense/vars": "model_00007.weights.h5",
353
+ "/layers/qwen_transformer_decoder_38/_self_attention_layer/_key_dense/vars": "model_00007.weights.h5",
354
+ "/layers/qwen_transformer_decoder_38/_self_attention_layer/_output_dense/vars": "model_00007.weights.h5",
355
+ "/layers/qwen_transformer_decoder_38/_self_attention_layer/_query_dense/vars": "model_00007.weights.h5",
356
+ "/layers/qwen_transformer_decoder_38/_self_attention_layer/_value_dense/vars": "model_00007.weights.h5",
357
+ "/layers/qwen_transformer_decoder_38/_self_attention_layernorm/vars": "model_00007.weights.h5",
358
+ "/layers/qwen_transformer_decoder_39/_feedforward_gate_dense/vars": "model_00007.weights.h5",
359
+ "/layers/qwen_transformer_decoder_39/_feedforward_intermediate_dense/vars": "model_00007.weights.h5",
360
+ "/layers/qwen_transformer_decoder_39/_feedforward_layernorm/vars": "model_00007.weights.h5",
361
+ "/layers/qwen_transformer_decoder_39/_feedforward_output_dense/vars": "model_00008.weights.h5",
362
+ "/layers/qwen_transformer_decoder_39/_self_attention_layer/_key_dense/vars": "model_00008.weights.h5",
363
+ "/layers/qwen_transformer_decoder_39/_self_attention_layer/_output_dense/vars": "model_00008.weights.h5",
364
+ "/layers/qwen_transformer_decoder_39/_self_attention_layer/_query_dense/vars": "model_00008.weights.h5",
365
+ "/layers/qwen_transformer_decoder_39/_self_attention_layer/_value_dense/vars": "model_00008.weights.h5",
366
+ "/layers/qwen_transformer_decoder_39/_self_attention_layernorm/vars": "model_00008.weights.h5",
367
+ "/layers/qwen_transformer_decoder_40/_feedforward_gate_dense/vars": "model_00008.weights.h5",
368
+ "/layers/qwen_transformer_decoder_40/_feedforward_intermediate_dense/vars": "model_00008.weights.h5",
369
+ "/layers/qwen_transformer_decoder_40/_feedforward_layernorm/vars": "model_00008.weights.h5",
370
+ "/layers/qwen_transformer_decoder_40/_feedforward_output_dense/vars": "model_00008.weights.h5",
371
+ "/layers/qwen_transformer_decoder_40/_self_attention_layer/_key_dense/vars": "model_00008.weights.h5",
372
+ "/layers/qwen_transformer_decoder_40/_self_attention_layer/_output_dense/vars": "model_00008.weights.h5",
373
+ "/layers/qwen_transformer_decoder_40/_self_attention_layer/_query_dense/vars": "model_00008.weights.h5",
374
+ "/layers/qwen_transformer_decoder_40/_self_attention_layer/_value_dense/vars": "model_00008.weights.h5",
375
+ "/layers/qwen_transformer_decoder_40/_self_attention_layernorm/vars": "model_00008.weights.h5",
376
+ "/layers/qwen_transformer_decoder_41/_feedforward_gate_dense/vars": "model_00008.weights.h5",
377
+ "/layers/qwen_transformer_decoder_41/_feedforward_intermediate_dense/vars": "model_00008.weights.h5",
378
+ "/layers/qwen_transformer_decoder_41/_feedforward_layernorm/vars": "model_00008.weights.h5",
379
+ "/layers/qwen_transformer_decoder_41/_feedforward_output_dense/vars": "model_00008.weights.h5",
380
+ "/layers/qwen_transformer_decoder_41/_self_attention_layer/_key_dense/vars": "model_00008.weights.h5",
381
+ "/layers/qwen_transformer_decoder_41/_self_attention_layer/_output_dense/vars": "model_00008.weights.h5",
382
+ "/layers/qwen_transformer_decoder_41/_self_attention_layer/_query_dense/vars": "model_00008.weights.h5",
383
+ "/layers/qwen_transformer_decoder_41/_self_attention_layer/_value_dense/vars": "model_00008.weights.h5",
384
+ "/layers/qwen_transformer_decoder_41/_self_attention_layernorm/vars": "model_00008.weights.h5",
385
+ "/layers/qwen_transformer_decoder_42/_feedforward_gate_dense/vars": "model_00008.weights.h5",
386
+ "/layers/qwen_transformer_decoder_42/_feedforward_intermediate_dense/vars": "model_00008.weights.h5",
387
+ "/layers/qwen_transformer_decoder_42/_feedforward_layernorm/vars": "model_00008.weights.h5",
388
+ "/layers/qwen_transformer_decoder_42/_feedforward_output_dense/vars": "model_00008.weights.h5",
389
+ "/layers/qwen_transformer_decoder_42/_self_attention_layer/_key_dense/vars": "model_00008.weights.h5",
390
+ "/layers/qwen_transformer_decoder_42/_self_attention_layer/_output_dense/vars": "model_00008.weights.h5",
391
+ "/layers/qwen_transformer_decoder_42/_self_attention_layer/_query_dense/vars": "model_00008.weights.h5",
392
+ "/layers/qwen_transformer_decoder_42/_self_attention_layer/_value_dense/vars": "model_00008.weights.h5",
393
+ "/layers/qwen_transformer_decoder_42/_self_attention_layernorm/vars": "model_00008.weights.h5",
394
+ "/layers/qwen_transformer_decoder_43/_feedforward_gate_dense/vars": "model_00008.weights.h5",
395
+ "/layers/qwen_transformer_decoder_43/_feedforward_intermediate_dense/vars": "model_00008.weights.h5",
396
+ "/layers/qwen_transformer_decoder_43/_feedforward_layernorm/vars": "model_00008.weights.h5",
397
+ "/layers/qwen_transformer_decoder_43/_feedforward_output_dense/vars": "model_00008.weights.h5",
398
+ "/layers/qwen_transformer_decoder_43/_self_attention_layer/_key_dense/vars": "model_00008.weights.h5",
399
+ "/layers/qwen_transformer_decoder_43/_self_attention_layer/_output_dense/vars": "model_00008.weights.h5",
400
+ "/layers/qwen_transformer_decoder_43/_self_attention_layer/_query_dense/vars": "model_00008.weights.h5",
401
+ "/layers/qwen_transformer_decoder_43/_self_attention_layer/_value_dense/vars": "model_00008.weights.h5",
402
+ "/layers/qwen_transformer_decoder_43/_self_attention_layernorm/vars": "model_00008.weights.h5",
403
+ "/layers/qwen_transformer_decoder_44/_feedforward_gate_dense/vars": "model_00008.weights.h5",
404
+ "/layers/qwen_transformer_decoder_44/_feedforward_intermediate_dense/vars": "model_00008.weights.h5",
405
+ "/layers/qwen_transformer_decoder_44/_feedforward_layernorm/vars": "model_00008.weights.h5",
406
+ "/layers/qwen_transformer_decoder_44/_feedforward_output_dense/vars": "model_00008.weights.h5",
407
+ "/layers/qwen_transformer_decoder_44/_self_attention_layer/_key_dense/vars": "model_00008.weights.h5",
408
+ "/layers/qwen_transformer_decoder_44/_self_attention_layer/_output_dense/vars": "model_00008.weights.h5",
409
+ "/layers/qwen_transformer_decoder_44/_self_attention_layer/_query_dense/vars": "model_00008.weights.h5",
410
+ "/layers/qwen_transformer_decoder_44/_self_attention_layer/_value_dense/vars": "model_00008.weights.h5",
411
+ "/layers/qwen_transformer_decoder_44/_self_attention_layernorm/vars": "model_00008.weights.h5",
412
+ "/layers/qwen_transformer_decoder_45/_feedforward_gate_dense/vars": "model_00009.weights.h5",
413
+ "/layers/qwen_transformer_decoder_45/_feedforward_intermediate_dense/vars": "model_00009.weights.h5",
414
+ "/layers/qwen_transformer_decoder_45/_feedforward_layernorm/vars": "model_00009.weights.h5",
415
+ "/layers/qwen_transformer_decoder_45/_feedforward_output_dense/vars": "model_00009.weights.h5",
416
+ "/layers/qwen_transformer_decoder_45/_self_attention_layer/_key_dense/vars": "model_00009.weights.h5",
417
+ "/layers/qwen_transformer_decoder_45/_self_attention_layer/_output_dense/vars": "model_00009.weights.h5",
418
+ "/layers/qwen_transformer_decoder_45/_self_attention_layer/_query_dense/vars": "model_00009.weights.h5",
419
+ "/layers/qwen_transformer_decoder_45/_self_attention_layer/_value_dense/vars": "model_00009.weights.h5",
420
+ "/layers/qwen_transformer_decoder_45/_self_attention_layernorm/vars": "model_00009.weights.h5",
421
+ "/layers/qwen_transformer_decoder_46/_feedforward_gate_dense/vars": "model_00009.weights.h5",
422
+ "/layers/qwen_transformer_decoder_46/_feedforward_intermediate_dense/vars": "model_00009.weights.h5",
423
+ "/layers/qwen_transformer_decoder_46/_feedforward_layernorm/vars": "model_00009.weights.h5",
424
+ "/layers/qwen_transformer_decoder_46/_feedforward_output_dense/vars": "model_00009.weights.h5",
425
+ "/layers/qwen_transformer_decoder_46/_self_attention_layer/_key_dense/vars": "model_00009.weights.h5",
426
+ "/layers/qwen_transformer_decoder_46/_self_attention_layer/_output_dense/vars": "model_00009.weights.h5",
427
+ "/layers/qwen_transformer_decoder_46/_self_attention_layer/_query_dense/vars": "model_00009.weights.h5",
428
+ "/layers/qwen_transformer_decoder_46/_self_attention_layer/_value_dense/vars": "model_00009.weights.h5",
429
+ "/layers/qwen_transformer_decoder_46/_self_attention_layernorm/vars": "model_00009.weights.h5",
430
+ "/layers/qwen_transformer_decoder_47/_feedforward_gate_dense/vars": "model_00009.weights.h5",
431
+ "/layers/qwen_transformer_decoder_47/_feedforward_intermediate_dense/vars": "model_00009.weights.h5",
432
+ "/layers/qwen_transformer_decoder_47/_feedforward_layernorm/vars": "model_00009.weights.h5",
433
+ "/layers/qwen_transformer_decoder_47/_feedforward_output_dense/vars": "model_00009.weights.h5",
434
+ "/layers/qwen_transformer_decoder_47/_self_attention_layer/_key_dense/vars": "model_00009.weights.h5",
435
+ "/layers/qwen_transformer_decoder_47/_self_attention_layer/_output_dense/vars": "model_00009.weights.h5",
436
+ "/layers/qwen_transformer_decoder_47/_self_attention_layer/_query_dense/vars": "model_00009.weights.h5",
437
+ "/layers/qwen_transformer_decoder_47/_self_attention_layer/_value_dense/vars": "model_00009.weights.h5",
438
+ "/layers/qwen_transformer_decoder_47/_self_attention_layernorm/vars": "model_00009.weights.h5",
439
+ "/layers/qwen_transformer_decoder_48/_feedforward_gate_dense/vars": "model_00009.weights.h5",
440
+ "/layers/qwen_transformer_decoder_48/_feedforward_intermediate_dense/vars": "model_00009.weights.h5",
441
+ "/layers/qwen_transformer_decoder_48/_feedforward_layernorm/vars": "model_00009.weights.h5",
442
+ "/layers/qwen_transformer_decoder_48/_feedforward_output_dense/vars": "model_00009.weights.h5",
443
+ "/layers/qwen_transformer_decoder_48/_self_attention_layer/_key_dense/vars": "model_00009.weights.h5",
444
+ "/layers/qwen_transformer_decoder_48/_self_attention_layer/_output_dense/vars": "model_00009.weights.h5",
445
+ "/layers/qwen_transformer_decoder_48/_self_attention_layer/_query_dense/vars": "model_00009.weights.h5",
446
+ "/layers/qwen_transformer_decoder_48/_self_attention_layer/_value_dense/vars": "model_00009.weights.h5",
447
+ "/layers/qwen_transformer_decoder_48/_self_attention_layernorm/vars": "model_00009.weights.h5",
448
+ "/layers/qwen_transformer_decoder_49/_feedforward_gate_dense/vars": "model_00009.weights.h5",
449
+ "/layers/qwen_transformer_decoder_49/_feedforward_intermediate_dense/vars": "model_00009.weights.h5",
450
+ "/layers/qwen_transformer_decoder_49/_feedforward_layernorm/vars": "model_00009.weights.h5",
451
+ "/layers/qwen_transformer_decoder_49/_feedforward_output_dense/vars": "model_00009.weights.h5",
452
+ "/layers/qwen_transformer_decoder_49/_self_attention_layer/_key_dense/vars": "model_00009.weights.h5",
453
+ "/layers/qwen_transformer_decoder_49/_self_attention_layer/_output_dense/vars": "model_00009.weights.h5",
454
+ "/layers/qwen_transformer_decoder_49/_self_attention_layer/_query_dense/vars": "model_00009.weights.h5",
455
+ "/layers/qwen_transformer_decoder_49/_self_attention_layer/_value_dense/vars": "model_00009.weights.h5",
456
+ "/layers/qwen_transformer_decoder_49/_self_attention_layernorm/vars": "model_00009.weights.h5",
457
+ "/layers/qwen_transformer_decoder_50/_feedforward_gate_dense/vars": "model_00009.weights.h5",
458
+ "/layers/qwen_transformer_decoder_50/_feedforward_intermediate_dense/vars": "model_00010.weights.h5",
459
+ "/layers/qwen_transformer_decoder_50/_feedforward_layernorm/vars": "model_00010.weights.h5",
460
+ "/layers/qwen_transformer_decoder_50/_feedforward_output_dense/vars": "model_00010.weights.h5",
461
+ "/layers/qwen_transformer_decoder_50/_self_attention_layer/_key_dense/vars": "model_00010.weights.h5",
462
+ "/layers/qwen_transformer_decoder_50/_self_attention_layer/_output_dense/vars": "model_00010.weights.h5",
463
+ "/layers/qwen_transformer_decoder_50/_self_attention_layer/_query_dense/vars": "model_00010.weights.h5",
464
+ "/layers/qwen_transformer_decoder_50/_self_attention_layer/_value_dense/vars": "model_00010.weights.h5",
465
+ "/layers/qwen_transformer_decoder_50/_self_attention_layernorm/vars": "model_00010.weights.h5",
466
+ "/layers/qwen_transformer_decoder_51/_feedforward_gate_dense/vars": "model_00010.weights.h5",
467
+ "/layers/qwen_transformer_decoder_51/_feedforward_intermediate_dense/vars": "model_00010.weights.h5",
468
+ "/layers/qwen_transformer_decoder_51/_feedforward_layernorm/vars": "model_00010.weights.h5",
469
+ "/layers/qwen_transformer_decoder_51/_feedforward_output_dense/vars": "model_00010.weights.h5",
470
+ "/layers/qwen_transformer_decoder_51/_self_attention_layer/_key_dense/vars": "model_00010.weights.h5",
471
+ "/layers/qwen_transformer_decoder_51/_self_attention_layer/_output_dense/vars": "model_00010.weights.h5",
472
+ "/layers/qwen_transformer_decoder_51/_self_attention_layer/_query_dense/vars": "model_00010.weights.h5",
473
+ "/layers/qwen_transformer_decoder_51/_self_attention_layer/_value_dense/vars": "model_00010.weights.h5",
474
+ "/layers/qwen_transformer_decoder_51/_self_attention_layernorm/vars": "model_00010.weights.h5",
475
+ "/layers/qwen_transformer_decoder_52/_feedforward_gate_dense/vars": "model_00010.weights.h5",
476
+ "/layers/qwen_transformer_decoder_52/_feedforward_intermediate_dense/vars": "model_00010.weights.h5",
477
+ "/layers/qwen_transformer_decoder_52/_feedforward_layernorm/vars": "model_00010.weights.h5",
478
+ "/layers/qwen_transformer_decoder_52/_feedforward_output_dense/vars": "model_00010.weights.h5",
479
+ "/layers/qwen_transformer_decoder_52/_self_attention_layer/_key_dense/vars": "model_00010.weights.h5",
480
+ "/layers/qwen_transformer_decoder_52/_self_attention_layer/_output_dense/vars": "model_00010.weights.h5",
481
+ "/layers/qwen_transformer_decoder_52/_self_attention_layer/_query_dense/vars": "model_00010.weights.h5",
482
+ "/layers/qwen_transformer_decoder_52/_self_attention_layer/_value_dense/vars": "model_00010.weights.h5",
483
+ "/layers/qwen_transformer_decoder_52/_self_attention_layernorm/vars": "model_00010.weights.h5",
484
+ "/layers/qwen_transformer_decoder_53/_feedforward_gate_dense/vars": "model_00010.weights.h5",
485
+ "/layers/qwen_transformer_decoder_53/_feedforward_intermediate_dense/vars": "model_00010.weights.h5",
486
+ "/layers/qwen_transformer_decoder_53/_feedforward_layernorm/vars": "model_00010.weights.h5",
487
+ "/layers/qwen_transformer_decoder_53/_feedforward_output_dense/vars": "model_00010.weights.h5",
488
+ "/layers/qwen_transformer_decoder_53/_self_attention_layer/_key_dense/vars": "model_00010.weights.h5",
489
+ "/layers/qwen_transformer_decoder_53/_self_attention_layer/_output_dense/vars": "model_00010.weights.h5",
490
+ "/layers/qwen_transformer_decoder_53/_self_attention_layer/_query_dense/vars": "model_00010.weights.h5",
491
+ "/layers/qwen_transformer_decoder_53/_self_attention_layer/_value_dense/vars": "model_00010.weights.h5",
492
+ "/layers/qwen_transformer_decoder_53/_self_attention_layernorm/vars": "model_00010.weights.h5",
493
+ "/layers/qwen_transformer_decoder_54/_feedforward_gate_dense/vars": "model_00010.weights.h5",
494
+ "/layers/qwen_transformer_decoder_54/_feedforward_intermediate_dense/vars": "model_00010.weights.h5",
495
+ "/layers/qwen_transformer_decoder_54/_feedforward_layernorm/vars": "model_00010.weights.h5",
496
+ "/layers/qwen_transformer_decoder_54/_feedforward_output_dense/vars": "model_00010.weights.h5",
497
+ "/layers/qwen_transformer_decoder_54/_self_attention_layer/_key_dense/vars": "model_00010.weights.h5",
498
+ "/layers/qwen_transformer_decoder_54/_self_attention_layer/_output_dense/vars": "model_00010.weights.h5",
499
+ "/layers/qwen_transformer_decoder_54/_self_attention_layer/_query_dense/vars": "model_00010.weights.h5",
500
+ "/layers/qwen_transformer_decoder_54/_self_attention_layer/_value_dense/vars": "model_00010.weights.h5",
501
+ "/layers/qwen_transformer_decoder_54/_self_attention_layernorm/vars": "model_00010.weights.h5",
502
+ "/layers/qwen_transformer_decoder_55/_feedforward_gate_dense/vars": "model_00010.weights.h5",
503
+ "/layers/qwen_transformer_decoder_55/_feedforward_intermediate_dense/vars": "model_00010.weights.h5",
504
+ "/layers/qwen_transformer_decoder_55/_feedforward_layernorm/vars": "model_00010.weights.h5",
505
+ "/layers/qwen_transformer_decoder_55/_feedforward_output_dense/vars": "model_00011.weights.h5",
506
+ "/layers/qwen_transformer_decoder_55/_self_attention_layer/_key_dense/vars": "model_00011.weights.h5",
507
+ "/layers/qwen_transformer_decoder_55/_self_attention_layer/_output_dense/vars": "model_00011.weights.h5",
508
+ "/layers/qwen_transformer_decoder_55/_self_attention_layer/_query_dense/vars": "model_00011.weights.h5",
509
+ "/layers/qwen_transformer_decoder_55/_self_attention_layer/_value_dense/vars": "model_00011.weights.h5",
510
+ "/layers/qwen_transformer_decoder_55/_self_attention_layernorm/vars": "model_00011.weights.h5",
511
+ "/layers/qwen_transformer_decoder_56/_feedforward_gate_dense/vars": "model_00011.weights.h5",
512
+ "/layers/qwen_transformer_decoder_56/_feedforward_intermediate_dense/vars": "model_00011.weights.h5",
513
+ "/layers/qwen_transformer_decoder_56/_feedforward_layernorm/vars": "model_00011.weights.h5",
514
+ "/layers/qwen_transformer_decoder_56/_feedforward_output_dense/vars": "model_00011.weights.h5",
515
+ "/layers/qwen_transformer_decoder_56/_self_attention_layer/_key_dense/vars": "model_00011.weights.h5",
516
+ "/layers/qwen_transformer_decoder_56/_self_attention_layer/_output_dense/vars": "model_00011.weights.h5",
517
+ "/layers/qwen_transformer_decoder_56/_self_attention_layer/_query_dense/vars": "model_00011.weights.h5",
518
+ "/layers/qwen_transformer_decoder_56/_self_attention_layer/_value_dense/vars": "model_00011.weights.h5",
519
+ "/layers/qwen_transformer_decoder_56/_self_attention_layernorm/vars": "model_00011.weights.h5",
520
+ "/layers/qwen_transformer_decoder_57/_feedforward_gate_dense/vars": "model_00011.weights.h5",
521
+ "/layers/qwen_transformer_decoder_57/_feedforward_intermediate_dense/vars": "model_00011.weights.h5",
522
+ "/layers/qwen_transformer_decoder_57/_feedforward_layernorm/vars": "model_00011.weights.h5",
523
+ "/layers/qwen_transformer_decoder_57/_feedforward_output_dense/vars": "model_00011.weights.h5",
524
+ "/layers/qwen_transformer_decoder_57/_self_attention_layer/_key_dense/vars": "model_00011.weights.h5",
525
+ "/layers/qwen_transformer_decoder_57/_self_attention_layer/_output_dense/vars": "model_00011.weights.h5",
526
+ "/layers/qwen_transformer_decoder_57/_self_attention_layer/_query_dense/vars": "model_00011.weights.h5",
527
+ "/layers/qwen_transformer_decoder_57/_self_attention_layer/_value_dense/vars": "model_00011.weights.h5",
528
+ "/layers/qwen_transformer_decoder_57/_self_attention_layernorm/vars": "model_00011.weights.h5",
529
+ "/layers/qwen_transformer_decoder_58/_feedforward_gate_dense/vars": "model_00011.weights.h5",
530
+ "/layers/qwen_transformer_decoder_58/_feedforward_intermediate_dense/vars": "model_00011.weights.h5",
531
+ "/layers/qwen_transformer_decoder_58/_feedforward_layernorm/vars": "model_00011.weights.h5",
532
+ "/layers/qwen_transformer_decoder_58/_feedforward_output_dense/vars": "model_00011.weights.h5",
533
+ "/layers/qwen_transformer_decoder_58/_self_attention_layer/_key_dense/vars": "model_00011.weights.h5",
534
+ "/layers/qwen_transformer_decoder_58/_self_attention_layer/_output_dense/vars": "model_00011.weights.h5",
535
+ "/layers/qwen_transformer_decoder_58/_self_attention_layer/_query_dense/vars": "model_00011.weights.h5",
536
+ "/layers/qwen_transformer_decoder_58/_self_attention_layer/_value_dense/vars": "model_00011.weights.h5",
537
+ "/layers/qwen_transformer_decoder_58/_self_attention_layernorm/vars": "model_00011.weights.h5",
538
+ "/layers/qwen_transformer_decoder_59/_feedforward_gate_dense/vars": "model_00011.weights.h5",
539
+ "/layers/qwen_transformer_decoder_59/_feedforward_intermediate_dense/vars": "model_00011.weights.h5",
540
+ "/layers/qwen_transformer_decoder_59/_feedforward_layernorm/vars": "model_00011.weights.h5",
541
+ "/layers/qwen_transformer_decoder_59/_feedforward_output_dense/vars": "model_00011.weights.h5",
542
+ "/layers/qwen_transformer_decoder_59/_self_attention_layer/_key_dense/vars": "model_00011.weights.h5",
543
+ "/layers/qwen_transformer_decoder_59/_self_attention_layer/_output_dense/vars": "model_00011.weights.h5",
544
+ "/layers/qwen_transformer_decoder_59/_self_attention_layer/_query_dense/vars": "model_00011.weights.h5",
545
+ "/layers/qwen_transformer_decoder_59/_self_attention_layer/_value_dense/vars": "model_00011.weights.h5",
546
+ "/layers/qwen_transformer_decoder_59/_self_attention_layernorm/vars": "model_00011.weights.h5",
547
+ "/layers/qwen_transformer_decoder_60/_feedforward_gate_dense/vars": "model_00011.weights.h5",
548
+ "/layers/qwen_transformer_decoder_60/_feedforward_intermediate_dense/vars": "model_00011.weights.h5",
549
+ "/layers/qwen_transformer_decoder_60/_feedforward_layernorm/vars": "model_00011.weights.h5",
550
+ "/layers/qwen_transformer_decoder_60/_feedforward_output_dense/vars": "model_00011.weights.h5",
551
+ "/layers/qwen_transformer_decoder_60/_self_attention_layer/_key_dense/vars": "model_00011.weights.h5",
552
+ "/layers/qwen_transformer_decoder_60/_self_attention_layer/_output_dense/vars": "model_00011.weights.h5",
553
+ "/layers/qwen_transformer_decoder_60/_self_attention_layer/_query_dense/vars": "model_00011.weights.h5",
554
+ "/layers/qwen_transformer_decoder_60/_self_attention_layer/_value_dense/vars": "model_00011.weights.h5",
555
+ "/layers/qwen_transformer_decoder_60/_self_attention_layernorm/vars": "model_00011.weights.h5",
556
+ "/layers/qwen_transformer_decoder_61/_feedforward_gate_dense/vars": "model_00012.weights.h5",
557
+ "/layers/qwen_transformer_decoder_61/_feedforward_intermediate_dense/vars": "model_00012.weights.h5",
558
+ "/layers/qwen_transformer_decoder_61/_feedforward_layernorm/vars": "model_00012.weights.h5",
559
+ "/layers/qwen_transformer_decoder_61/_feedforward_output_dense/vars": "model_00012.weights.h5",
560
+ "/layers/qwen_transformer_decoder_61/_self_attention_layer/_key_dense/vars": "model_00012.weights.h5",
561
+ "/layers/qwen_transformer_decoder_61/_self_attention_layer/_output_dense/vars": "model_00012.weights.h5",
562
+ "/layers/qwen_transformer_decoder_61/_self_attention_layer/_query_dense/vars": "model_00012.weights.h5",
563
+ "/layers/qwen_transformer_decoder_61/_self_attention_layer/_value_dense/vars": "model_00012.weights.h5",
564
+ "/layers/qwen_transformer_decoder_61/_self_attention_layernorm/vars": "model_00012.weights.h5",
565
+ "/layers/qwen_transformer_decoder_62/_feedforward_gate_dense/vars": "model_00012.weights.h5",
566
+ "/layers/qwen_transformer_decoder_62/_feedforward_intermediate_dense/vars": "model_00012.weights.h5",
567
+ "/layers/qwen_transformer_decoder_62/_feedforward_layernorm/vars": "model_00012.weights.h5",
568
+ "/layers/qwen_transformer_decoder_62/_feedforward_output_dense/vars": "model_00012.weights.h5",
569
+ "/layers/qwen_transformer_decoder_62/_self_attention_layer/_key_dense/vars": "model_00012.weights.h5",
570
+ "/layers/qwen_transformer_decoder_62/_self_attention_layer/_output_dense/vars": "model_00012.weights.h5",
571
+ "/layers/qwen_transformer_decoder_62/_self_attention_layer/_query_dense/vars": "model_00012.weights.h5",
572
+ "/layers/qwen_transformer_decoder_62/_self_attention_layer/_value_dense/vars": "model_00012.weights.h5",
573
+ "/layers/qwen_transformer_decoder_62/_self_attention_layernorm/vars": "model_00012.weights.h5",
574
+ "/layers/qwen_transformer_decoder_63/_feedforward_gate_dense/vars": "model_00012.weights.h5",
575
+ "/layers/qwen_transformer_decoder_63/_feedforward_intermediate_dense/vars": "model_00012.weights.h5",
576
+ "/layers/qwen_transformer_decoder_63/_feedforward_layernorm/vars": "model_00012.weights.h5",
577
+ "/layers/qwen_transformer_decoder_63/_feedforward_output_dense/vars": "model_00012.weights.h5",
578
+ "/layers/qwen_transformer_decoder_63/_self_attention_layer/_key_dense/vars": "model_00012.weights.h5",
579
+ "/layers/qwen_transformer_decoder_63/_self_attention_layer/_output_dense/vars": "model_00012.weights.h5",
580
+ "/layers/qwen_transformer_decoder_63/_self_attention_layer/_query_dense/vars": "model_00012.weights.h5",
581
+ "/layers/qwen_transformer_decoder_63/_self_attention_layer/_value_dense/vars": "model_00012.weights.h5",
582
+ "/layers/qwen_transformer_decoder_63/_self_attention_layernorm/vars": "model_00012.weights.h5",
583
+ "/layers/qwen_layer_norm/vars": "model_00012.weights.h5"
584
+ }
585
+ }
model_00000.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f6adc22b7bf443160aaa45a1e00ea42e9202b9777ca55f06e46fb318a40683cf
3
+ size 10695702368
model_00001.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a7e0aaa83ddd2b9b0eb3cf5afc6842aada11a62f52bba890517812b3cb4ccf71
3
+ size 10318541312
model_00002.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7f4f1d91e8b7423e0bafb04ad1868b48d59ea513b60a89869f4304152d77d958
3
+ size 10570250088
model_00003.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1a0a9ee38fcd85a9c7b27d0f5ce51f9151c145a8dceaea2d374eaa75b63f59b9
3
+ size 10318518520
model_00004.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6c081de853f2fbab053a60ff97ab51f7304b12c7c992bca4415916bb73146108
3
+ size 10318541312
model_00005.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ac02e1a21e29b495bdddfe00bf240e019dcef103bd34c1afa7f031402e54da54
3
+ size 10570250088
model_00006.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a5a2474d22bd2bf52b3dd9cf100a2645eaa1bd51e73a795f9cfd758b8b641585
3
+ size 10318518520
model_00007.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:00d2adc79be8c4243583a22cda64d218c56fe2b476d4f287d9b761d84781d64c
3
+ size 10318541312
model_00008.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a3880e70a4b47f3e74cbb3466cb81a23dbda55ee3bd9db7b21a138a6929246b6
3
+ size 10570250088
model_00009.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a2f575e613e1e01a3b94e22edb636c93f3768ba6fdcd6a74e4ae52b6f0fef394
3
+ size 10318518520
model_00010.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:59b790abc1e6055ced4b5d0bb869557d8322d9f797f0846df4df045c63fd817c
3
+ size 10318541312
model_00011.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0976c95261a94e3d1f3b1f9e3f27e856f50e48fce54e0af33283e68af59b5fe9
3
+ size 10570250088
model_00012.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:772f1898019131c4053d32090a3bf0c593ced2091aa60f973416806fe1a6a2b2
3
+ size 5851394288
preprocessor.json ADDED
@@ -0,0 +1,65 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "module": "keras_hub.src.models.qwen.qwen_causal_lm_preprocessor",
3
+ "class_name": "QwenCausalLMPreprocessor",
4
+ "config": {
5
+ "name": "qwen_causal_lm_preprocessor_2",
6
+ "trainable": true,
7
+ "dtype": {
8
+ "module": "keras",
9
+ "class_name": "DTypePolicy",
10
+ "config": {
11
+ "name": "float32"
12
+ },
13
+ "registered_name": null
14
+ },
15
+ "tokenizer": {
16
+ "module": "keras_hub.src.models.qwen.qwen_tokenizer",
17
+ "class_name": "QwenTokenizer",
18
+ "config": {
19
+ "name": "qwen_tokenizer",
20
+ "trainable": true,
21
+ "dtype": {
22
+ "module": "keras",
23
+ "class_name": "DTypePolicy",
24
+ "config": {
25
+ "name": "int32"
26
+ },
27
+ "registered_name": null
28
+ },
29
+ "config_file": "tokenizer.json",
30
+ "sequence_length": null,
31
+ "add_prefix_space": false,
32
+ "unsplittable_tokens": [
33
+ "<|repo_name|>",
34
+ "<|im_end|>",
35
+ "<|file_sep|>",
36
+ "<|video_pad|>",
37
+ "<|quad_end|>",
38
+ "<|fim_prefix|>",
39
+ "<|box_end|>",
40
+ "</tool_call>",
41
+ "<|vision_end|>",
42
+ "<|endoftext|>",
43
+ "<|object_ref_start|>",
44
+ "<|box_start|>",
45
+ "<|object_ref_end|>",
46
+ "<|vision_start|>",
47
+ "<|image_pad|>",
48
+ "<|fim_suffix|>",
49
+ "<tool_call>",
50
+ "<|quad_start|>",
51
+ "<|im_start|>",
52
+ "<|vision_pad|>",
53
+ "<|fim_middle|>",
54
+ "<|fim_pad|>"
55
+ ]
56
+ },
57
+ "registered_name": "keras_hub>QwenTokenizer"
58
+ },
59
+ "config_file": "preprocessor.json",
60
+ "sequence_length": 1024,
61
+ "add_start_token": true,
62
+ "add_end_token": true
63
+ },
64
+ "registered_name": "keras_hub>QwenCausalLMPreprocessor"
65
+ }
task.json ADDED
@@ -0,0 +1,95 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "module": "keras_hub.src.models.qwen.qwen_causal_lm",
3
+ "class_name": "QwenCausalLM",
4
+ "config": {
5
+ "backbone": {
6
+ "module": "keras_hub.src.models.qwen.qwen_backbone",
7
+ "class_name": "QwenBackbone",
8
+ "config": {
9
+ "name": "qwen_backbone",
10
+ "trainable": true,
11
+ "vocabulary_size": 152064,
12
+ "num_layers": 64,
13
+ "num_query_heads": 40,
14
+ "hidden_dim": 5120,
15
+ "intermediate_dim": 27648,
16
+ "rope_max_wavelength": 1000000.0,
17
+ "rope_scaling_factor": 1.0,
18
+ "num_key_value_heads": 8,
19
+ "layer_norm_epsilon": 1e-06,
20
+ "dropout": 0,
21
+ "tie_word_embeddings": false,
22
+ "use_sliding_window_attention": false,
23
+ "sliding_window_size": 131072
24
+ },
25
+ "registered_name": "keras_hub>QwenBackbone"
26
+ },
27
+ "preprocessor": {
28
+ "module": "keras_hub.src.models.qwen.qwen_causal_lm_preprocessor",
29
+ "class_name": "QwenCausalLMPreprocessor",
30
+ "config": {
31
+ "name": "qwen_causal_lm_preprocessor_2",
32
+ "trainable": true,
33
+ "dtype": {
34
+ "module": "keras",
35
+ "class_name": "DTypePolicy",
36
+ "config": {
37
+ "name": "float32"
38
+ },
39
+ "registered_name": null
40
+ },
41
+ "tokenizer": {
42
+ "module": "keras_hub.src.models.qwen.qwen_tokenizer",
43
+ "class_name": "QwenTokenizer",
44
+ "config": {
45
+ "name": "qwen_tokenizer",
46
+ "trainable": true,
47
+ "dtype": {
48
+ "module": "keras",
49
+ "class_name": "DTypePolicy",
50
+ "config": {
51
+ "name": "int32"
52
+ },
53
+ "registered_name": null
54
+ },
55
+ "config_file": "tokenizer.json",
56
+ "sequence_length": null,
57
+ "add_prefix_space": false,
58
+ "unsplittable_tokens": [
59
+ "<|repo_name|>",
60
+ "<|im_end|>",
61
+ "<|file_sep|>",
62
+ "<|video_pad|>",
63
+ "<|quad_end|>",
64
+ "<|fim_prefix|>",
65
+ "<|box_end|>",
66
+ "</tool_call>",
67
+ "<|vision_end|>",
68
+ "<|endoftext|>",
69
+ "<|object_ref_start|>",
70
+ "<|box_start|>",
71
+ "<|object_ref_end|>",
72
+ "<|vision_start|>",
73
+ "<|image_pad|>",
74
+ "<|fim_suffix|>",
75
+ "<tool_call>",
76
+ "<|quad_start|>",
77
+ "<|im_start|>",
78
+ "<|vision_pad|>",
79
+ "<|fim_middle|>",
80
+ "<|fim_pad|>"
81
+ ]
82
+ },
83
+ "registered_name": "keras_hub>QwenTokenizer"
84
+ },
85
+ "config_file": "preprocessor.json",
86
+ "sequence_length": 1024,
87
+ "add_start_token": true,
88
+ "add_end_token": true
89
+ },
90
+ "registered_name": "keras_hub>QwenCausalLMPreprocessor"
91
+ },
92
+ "name": "qwen_causal_lm"
93
+ },
94
+ "registered_name": "keras_hub>QwenCausalLM"
95
+ }
tokenizer.json ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "module": "keras_hub.src.models.qwen.qwen_tokenizer",
3
+ "class_name": "QwenTokenizer",
4
+ "config": {
5
+ "name": "qwen_tokenizer",
6
+ "trainable": true,
7
+ "dtype": {
8
+ "module": "keras",
9
+ "class_name": "DTypePolicy",
10
+ "config": {
11
+ "name": "int32"
12
+ },
13
+ "registered_name": null
14
+ },
15
+ "config_file": "tokenizer.json",
16
+ "sequence_length": null,
17
+ "add_prefix_space": false,
18
+ "unsplittable_tokens": [
19
+ "<|repo_name|>",
20
+ "<|im_end|>",
21
+ "<|file_sep|>",
22
+ "<|video_pad|>",
23
+ "<|quad_end|>",
24
+ "<|fim_prefix|>",
25
+ "<|box_end|>",
26
+ "</tool_call>",
27
+ "<|vision_end|>",
28
+ "<|endoftext|>",
29
+ "<|object_ref_start|>",
30
+ "<|box_start|>",
31
+ "<|object_ref_end|>",
32
+ "<|vision_start|>",
33
+ "<|image_pad|>",
34
+ "<|fim_suffix|>",
35
+ "<tool_call>",
36
+ "<|quad_start|>",
37
+ "<|im_start|>",
38
+ "<|vision_pad|>",
39
+ "<|fim_middle|>",
40
+ "<|fim_pad|>"
41
+ ]
42
+ },
43
+ "registered_name": "keras_hub>QwenTokenizer"
44
+ }