Hugging Face's logo

keras
/
qwen3_embedding_0.6b_en

Text Generation
KerasHub
Model card Files
xet
 Community
prasadsachin commited on
Commit
fdaeab0
·
verified ·
1 Parent(s): 5d5a15d

Upload folder using huggingface_hub

Browse files
Files changed (9) hide show
  1. README.md +26 -0
  2. assets/tokenizer/merges.txt +0 -0
  3. assets/tokenizer/vocabulary.json +0 -0
  4. config.json +30 -0
  5. metadata.json +9 -0
  6. model.weights.h5 +3 -0
  7. preprocessor.json +69 -0
  8. task.json +107 -0
  9. tokenizer.json +48 -0
README.md ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: keras-hub
3
+ pipeline_tag: text-generation
4
+ ---
5
+ This is a [`Qwen3` model](https://keras.io/api/keras_hub/models/qwen3) 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:** qwen3_backbone
10
+ * **trainable:** True
11
+ * **dtype:** {'module': 'keras', 'class_name': 'DTypePolicy', 'config': {'name': 'float32'}, 'registered_name': None}
12
+ * **vocabulary_size:** 151669
13
+ * **num_layers:** 28
14
+ * **num_query_heads:** 16
15
+ * **hidden_dim:** 1024
16
+ * **head_dim:** 128
17
+ * **intermediate_dim:** 3072
18
+ * **rope_max_wavelength:** 1000000
19
+ * **rope_scaling_factor:** 1.0
20
+ * **num_key_value_heads:** 8
21
+ * **layer_norm_epsilon:** 1e-06
22
+ * **dropout:** 0.0
23
+ * **tie_word_embeddings:** True
24
+ * **sliding_window_size:** None
25
+
26
+ 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 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "module": "keras_hub.src.models.qwen3.qwen3_backbone",
3
+ "class_name": "Qwen3Backbone",
4
+ "config": {
5
+ "name": "qwen3_backbone",
6
+ "trainable": true,
7
+ "dtype": {
8
+ "module": "keras",
9
+ "class_name": "DTypePolicy",
10
+ "config": {
11
+ "name": "float32"
12
+ },
13
+ "registered_name": null
14
+ },
15
+ "vocabulary_size": 151669,
16
+ "num_layers": 28,
17
+ "num_query_heads": 16,
18
+ "hidden_dim": 1024,
19
+ "head_dim": 128,
20
+ "intermediate_dim": 3072,
21
+ "rope_max_wavelength": 1000000,
22
+ "rope_scaling_factor": 1.0,
23
+ "num_key_value_heads": 8,
24
+ "layer_norm_epsilon": 1e-06,
25
+ "dropout": 0.0,
26
+ "tie_word_embeddings": true,
27
+ "sliding_window_size": null
28
+ },
29
+ "registered_name": "keras_hub>Qwen3Backbone"
30
+ }
metadata.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "keras_version": "3.11.3",
3
+ "keras_hub_version": "0.23.0.dev0",
4
+ "parameter_count": 595776512,
5
+ "date_saved": "2025-11-14@00:06:25",
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:faf25ea60a1b08fbc5e03d894c0c4c18793868f3dafd4d71c1ea4cbe1b575d2c
3
+ size 2384223192
preprocessor.json ADDED
@@ -0,0 +1,69 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "module": "keras_hub.src.models.qwen3.qwen3_causal_lm_preprocessor",
3
+ "class_name": "Qwen3CausalLMPreprocessor",
4
+ "config": {
5
+ "name": "qwen3_causal_lm_preprocessor_1",
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.qwen3.qwen3_tokenizer",
17
+ "class_name": "Qwen3Tokenizer",
18
+ "config": {
19
+ "name": "qwen3_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
+ "<|endoftext|>",
34
+ "<|im_start|>",
35
+ "<tool_call>",
36
+ "<|box_start|>",
37
+ "<|object_ref_start|>",
38
+ "<|vision_end|>",
39
+ "<|vision_start|>",
40
+ "<|im_end|>",
41
+ "<|fim_suffix|>",
42
+ "<|quad_start|>",
43
+ "</think>",
44
+ "<|vision_pad|>",
45
+ "<|fim_middle|>",
46
+ "</tool_call>",
47
+ "<|image_pad|>",
48
+ "<|repo_name|>",
49
+ "<think>",
50
+ "<tool_response>",
51
+ "<|fim_pad|>",
52
+ "<|quad_end|>",
53
+ "<|object_ref_end|>",
54
+ "<|file_sep|>",
55
+ "<|box_end|>",
56
+ "</tool_response>",
57
+ "<|video_pad|>",
58
+ "<|fim_prefix|>"
59
+ ]
60
+ },
61
+ "registered_name": "keras_hub>Qwen3Tokenizer"
62
+ },
63
+ "config_file": "preprocessor.json",
64
+ "sequence_length": 1024,
65
+ "add_start_token": true,
66
+ "add_end_token": true
67
+ },
68
+ "registered_name": "keras_hub>Qwen3CausalLMPreprocessor"
69
+ }
task.json ADDED
@@ -0,0 +1,107 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "module": "keras_hub.src.models.qwen3.qwen3_causal_lm",
3
+ "class_name": "Qwen3CausalLM",
4
+ "config": {
5
+ "backbone": {
6
+ "module": "keras_hub.src.models.qwen3.qwen3_backbone",
7
+ "class_name": "Qwen3Backbone",
8
+ "config": {
9
+ "name": "qwen3_backbone",
10
+ "trainable": true,
11
+ "dtype": {
12
+ "module": "keras",
13
+ "class_name": "DTypePolicy",
14
+ "config": {
15
+ "name": "float32"
16
+ },
17
+ "registered_name": null
18
+ },
19
+ "vocabulary_size": 151669,
20
+ "num_layers": 28,
21
+ "num_query_heads": 16,
22
+ "hidden_dim": 1024,
23
+ "head_dim": 128,
24
+ "intermediate_dim": 3072,
25
+ "rope_max_wavelength": 1000000,
26
+ "rope_scaling_factor": 1.0,
27
+ "num_key_value_heads": 8,
28
+ "layer_norm_epsilon": 1e-06,
29
+ "dropout": 0.0,
30
+ "tie_word_embeddings": true,
31
+ "sliding_window_size": null
32
+ },
33
+ "registered_name": "keras_hub>Qwen3Backbone"
34
+ },
35
+ "preprocessor": {
36
+ "module": "keras_hub.src.models.qwen3.qwen3_causal_lm_preprocessor",
37
+ "class_name": "Qwen3CausalLMPreprocessor",
38
+ "config": {
39
+ "name": "qwen3_causal_lm_preprocessor_1",
40
+ "trainable": true,
41
+ "dtype": {
42
+ "module": "keras",
43
+ "class_name": "DTypePolicy",
44
+ "config": {
45
+ "name": "float32"
46
+ },
47
+ "registered_name": null
48
+ },
49
+ "tokenizer": {
50
+ "module": "keras_hub.src.models.qwen3.qwen3_tokenizer",
51
+ "class_name": "Qwen3Tokenizer",
52
+ "config": {
53
+ "name": "qwen3_tokenizer",
54
+ "trainable": true,
55
+ "dtype": {
56
+ "module": "keras",
57
+ "class_name": "DTypePolicy",
58
+ "config": {
59
+ "name": "int32"
60
+ },
61
+ "registered_name": null
62
+ },
63
+ "config_file": "tokenizer.json",
64
+ "sequence_length": null,
65
+ "add_prefix_space": false,
66
+ "unsplittable_tokens": [
67
+ "<|endoftext|>",
68
+ "<|im_start|>",
69
+ "<tool_call>",
70
+ "<|box_start|>",
71
+ "<|object_ref_start|>",
72
+ "<|vision_end|>",
73
+ "<|vision_start|>",
74
+ "<|im_end|>",
75
+ "<|fim_suffix|>",
76
+ "<|quad_start|>",
77
+ "</think>",
78
+ "<|vision_pad|>",
79
+ "<|fim_middle|>",
80
+ "</tool_call>",
81
+ "<|image_pad|>",
82
+ "<|repo_name|>",
83
+ "<think>",
84
+ "<tool_response>",
85
+ "<|fim_pad|>",
86
+ "<|quad_end|>",
87
+ "<|object_ref_end|>",
88
+ "<|file_sep|>",
89
+ "<|box_end|>",
90
+ "</tool_response>",
91
+ "<|video_pad|>",
92
+ "<|fim_prefix|>"
93
+ ]
94
+ },
95
+ "registered_name": "keras_hub>Qwen3Tokenizer"
96
+ },
97
+ "config_file": "preprocessor.json",
98
+ "sequence_length": 1024,
99
+ "add_start_token": true,
100
+ "add_end_token": true
101
+ },
102
+ "registered_name": "keras_hub>Qwen3CausalLMPreprocessor"
103
+ },
104
+ "name": "qwen3_causal_lm"
105
+ },
106
+ "registered_name": "keras_hub>Qwen3CausalLM"
107
+ }
tokenizer.json ADDED
@@ -0,0 +1,48 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "module": "keras_hub.src.models.qwen3.qwen3_tokenizer",
3
+ "class_name": "Qwen3Tokenizer",
4
+ "config": {
5
+ "name": "qwen3_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
+ "<|endoftext|>",
20
+ "<|im_start|>",
21
+ "<tool_call>",
22
+ "<|box_start|>",
23
+ "<|object_ref_start|>",
24
+ "<|vision_end|>",
25
+ "<|vision_start|>",
26
+ "<|im_end|>",
27
+ "<|fim_suffix|>",
28
+ "<|quad_start|>",
29
+ "</think>",
30
+ "<|vision_pad|>",
31
+ "<|fim_middle|>",
32
+ "</tool_call>",
33
+ "<|image_pad|>",
34
+ "<|repo_name|>",
35
+ "<think>",
36
+ "<tool_response>",
37
+ "<|fim_pad|>",
38
+ "<|quad_end|>",
39
+ "<|object_ref_end|>",
40
+ "<|file_sep|>",
41
+ "<|box_end|>",
42
+ "</tool_response>",
43
+ "<|video_pad|>",
44
+ "<|fim_prefix|>"
45
+ ]
46
+ },
47
+ "registered_name": "keras_hub>Qwen3Tokenizer"
48
+ }