Hugging Face's logo

keras
/
qwen2.5_3b_en

Text Generation
KerasHub
Model card Files
xet
Community
Divyasreepat commited on
Commit
00fd396
·
verified ·
1 Parent(s): 84794db

Upload folder using huggingface_hub

Browse files
Files changed (9) hide show
  1. README.md +25 -0
  2. assets/tokenizer/merges.txt +0 -0
  3. assets/tokenizer/vocabulary.json +0 -0
  4. config.json +24 -0
  5. metadata.json +9 -0
  6. model.weights.h5 +3 -0
  7. preprocessor.json +65 -0
  8. task.json +97 -0
  9. 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:** 151936
12
+ * **num_layers:** 36
13
+ * **num_query_heads:** 16
14
+ * **hidden_dim:** 2048
15
+ * **intermediate_dim:** 11008
16
+ * **rope_max_wavelength:** 1000000.0
17
+ * **rope_scaling_factor:** 1.0
18
+ * **num_key_value_heads:** 2
19
+ * **layer_norm_epsilon:** 1e-06
20
+ * **dropout:** 0
21
+ * **tie_word_embeddings:** True
22
+ * **use_sliding_window_attention:** [False]
23
+ * **sliding_window_size:** 32768
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 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": 151936,
8
+ "num_layers": 36,
9
+ "num_query_heads": 16,
10
+ "hidden_dim": 2048,
11
+ "intermediate_dim": 11008,
12
+ "rope_max_wavelength": 1000000.0,
13
+ "rope_scaling_factor": 1.0,
14
+ "num_key_value_heads": 2,
15
+ "layer_norm_epsilon": 1e-06,
16
+ "dropout": 0,
17
+ "tie_word_embeddings": true,
18
+ "use_sliding_window_attention": [
19
+ false
20
+ ],
21
+ "sliding_window_size": 32768
22
+ },
23
+ "registered_name": "keras_hub>QwenBackbone"
24
+ }
metadata.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "keras_version": "3.8.0",
3
+ "keras_hub_version": "0.20.0.dev202503250356",
4
+ "parameter_count": 3085938688,
5
+ "date_saved": "2025-03-28@12:59:33",
6
+ "tasks": [
7
+ "CausalLM"
8
+ ]
9
+ }
model.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:783ac97507ea215b3d846803d6605911c292613f9d83d55cffcc55a54fe9b142
3
+ size 12345018464
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
+ "<|im_start|>",
34
+ "<|im_end|>",
35
+ "<|object_ref_end|>",
36
+ "<|endoftext|>",
37
+ "<tool_call>",
38
+ "<|fim_middle|>",
39
+ "<|video_pad|>",
40
+ "<|box_end|>",
41
+ "<|fim_suffix|>",
42
+ "<|repo_name|>",
43
+ "<|object_ref_start|>",
44
+ "<|image_pad|>",
45
+ "<|box_start|>",
46
+ "<|quad_end|>",
47
+ "<|quad_start|>",
48
+ "<|vision_end|>",
49
+ "<|fim_prefix|>",
50
+ "</tool_call>",
51
+ "<|vision_pad|>",
52
+ "<|vision_start|>",
53
+ "<|fim_pad|>",
54
+ "<|file_sep|>"
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 +1,97 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": 151936,
12
+ "num_layers": 36,
13
+ "num_query_heads": 16,
14
+ "hidden_dim": 2048,
15
+ "intermediate_dim": 11008,
16
+ "rope_max_wavelength": 1000000.0,
17
+ "rope_scaling_factor": 1.0,
18
+ "num_key_value_heads": 2,
19
+ "layer_norm_epsilon": 1e-06,
20
+ "dropout": 0,
21
+ "tie_word_embeddings": true,
22
+ "use_sliding_window_attention": [
23
+ false
24
+ ],
25
+ "sliding_window_size": 32768
26
+ },
27
+ "registered_name": "keras_hub>QwenBackbone"
28
+ },
29
+ "preprocessor": {
30
+ "module": "keras_hub.src.models.qwen.qwen_causal_lm_preprocessor",
31
+ "class_name": "QwenCausalLMPreprocessor",
32
+ "config": {
33
+ "name": "qwen_causal_lm_preprocessor_2",
34
+ "trainable": true,
35
+ "dtype": {
36
+ "module": "keras",
37
+ "class_name": "DTypePolicy",
38
+ "config": {
39
+ "name": "float32"
40
+ },
41
+ "registered_name": null
42
+ },
43
+ "tokenizer": {
44
+ "module": "keras_hub.src.models.qwen.qwen_tokenizer",
45
+ "class_name": "QwenTokenizer",
46
+ "config": {
47
+ "name": "qwen_tokenizer",
48
+ "trainable": true,
49
+ "dtype": {
50
+ "module": "keras",
51
+ "class_name": "DTypePolicy",
52
+ "config": {
53
+ "name": "int32"
54
+ },
55
+ "registered_name": null
56
+ },
57
+ "config_file": "tokenizer.json",
58
+ "sequence_length": null,
59
+ "add_prefix_space": false,
60
+ "unsplittable_tokens": [
61
+ "<|im_start|>",
62
+ "<|im_end|>",
63
+ "<|object_ref_end|>",
64
+ "<|endoftext|>",
65
+ "<tool_call>",
66
+ "<|fim_middle|>",
67
+ "<|video_pad|>",
68
+ "<|box_end|>",
69
+ "<|fim_suffix|>",
70
+ "<|repo_name|>",
71
+ "<|object_ref_start|>",
72
+ "<|image_pad|>",
73
+ "<|box_start|>",
74
+ "<|quad_end|>",
75
+ "<|quad_start|>",
76
+ "<|vision_end|>",
77
+ "<|fim_prefix|>",
78
+ "</tool_call>",
79
+ "<|vision_pad|>",
80
+ "<|vision_start|>",
81
+ "<|fim_pad|>",
82
+ "<|file_sep|>"
83
+ ]
84
+ },
85
+ "registered_name": "keras_hub>QwenTokenizer"
86
+ },
87
+ "config_file": "preprocessor.json",
88
+ "sequence_length": 1024,
89
+ "add_start_token": true,
90
+ "add_end_token": true
91
+ },
92
+ "registered_name": "keras_hub>QwenCausalLMPreprocessor"
93
+ },
94
+ "name": "qwen_causal_lm"
95
+ },
96
+ "registered_name": "keras_hub>QwenCausalLM"
97
+ }
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
+ "<|im_start|>",
20
+ "<|im_end|>",
21
+ "<|object_ref_end|>",
22
+ "<|endoftext|>",
23
+ "<tool_call>",
24
+ "<|fim_middle|>",
25
+ "<|video_pad|>",
26
+ "<|box_end|>",
27
+ "<|fim_suffix|>",
28
+ "<|repo_name|>",
29
+ "<|object_ref_start|>",
30
+ "<|image_pad|>",
31
+ "<|box_start|>",
32
+ "<|quad_end|>",
33
+ "<|quad_start|>",
34
+ "<|vision_end|>",
35
+ "<|fim_prefix|>",
36
+ "</tool_call>",
37
+ "<|vision_pad|>",
38
+ "<|vision_start|>",
39
+ "<|fim_pad|>",
40
+ "<|file_sep|>"
41
+ ]
42
+ },
43
+ "registered_name": "keras_hub>QwenTokenizer"
44
+ }