Hugging Face's logo

keras
/
llama3.2_instruct_1b

Text Generation
KerasHub
Model card Files Community
Divyasreepat commited on
Commit
e803a0b
·
verified ·
1 Parent(s): f426132

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 +23 -0
  5. metadata.json +6 -0
  6. model.weights.h5 +3 -0
  7. preprocessor.json +51 -0
  8. task.json +82 -0
  9. tokenizer.json +30 -0
README.md ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: keras-hub
3
+ pipeline_tag: text-generation
4
+ ---
5
+ This is a [`Llama3` model](https://keras.io/api/keras_hub/models/llama3) 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:** llama3_backbone
10
+ * **trainable:** True
11
+ * **vocabulary_size:** 128256
12
+ * **num_layers:** 16
13
+ * **num_query_heads:** 32
14
+ * **hidden_dim:** 2048
15
+ * **intermediate_dim:** 8192
16
+ * **rope_max_wavelength:** 500000.0
17
+ * **rope_position_scaling_factor:** 1
18
+ * **rope_frequency_adjustment_factor:** 32
19
+ * **rope_low_freq_factor:** 1
20
+ * **rope_high_freq_factor:** 4
21
+ * **rope_pretraining_sequence_length:** 8192
22
+ * **num_key_value_heads:** 8
23
+ * **layer_norm_epsilon:** 1e-05
24
+ * **dropout:** 0
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 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "module": "keras_hub.src.models.llama3.llama3_backbone",
3
+ "class_name": "Llama3Backbone",
4
+ "config": {
5
+ "name": "llama3_backbone",
6
+ "trainable": true,
7
+ "vocabulary_size": 128256,
8
+ "num_layers": 16,
9
+ "num_query_heads": 32,
10
+ "hidden_dim": 2048,
11
+ "intermediate_dim": 8192,
12
+ "rope_max_wavelength": 500000.0,
13
+ "rope_position_scaling_factor": 1,
14
+ "rope_frequency_adjustment_factor": 32,
15
+ "rope_low_freq_factor": 1,
16
+ "rope_high_freq_factor": 4,
17
+ "rope_pretraining_sequence_length": 8192,
18
+ "num_key_value_heads": 8,
19
+ "layer_norm_epsilon": 1e-05,
20
+ "dropout": 0
21
+ },
22
+ "registered_name": "keras_hub>Llama3Backbone"
23
+ }
metadata.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "keras_version": "3.8.0",
3
+ "keras_hub_version": "0.18.1",
4
+ "parameter_count": 1498482688,
5
+ "date_saved": "2025-04-12@04:02:49"
6
+ }
model.weights.h5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:36cd1e64c1bccd85b2ab39676571657e4918bb1f5fa933cff8a1a95f2da264d9
3
+ size 2997527440
preprocessor.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "module": "keras_hub.src.models.llama3.llama3_causal_lm_preprocessor",
3
+ "class_name": "Llama3CausalLMPreprocessor",
4
+ "config": {
5
+ "name": "llama3_causal_lm_preprocessor",
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.llama3.llama3_tokenizer",
17
+ "class_name": "Llama3Tokenizer",
18
+ "config": {
19
+ "name": "llama3_tokenizer_1",
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
+ "<|begin_of_text|>",
34
+ "<|finetune_right_pad_id|>",
35
+ "<|start_header_id|>",
36
+ "<|end_of_text|>",
37
+ "<|python_tag|>",
38
+ "<|eot_id|>",
39
+ "<|eom_id|>",
40
+ "<|end_header_id|>"
41
+ ]
42
+ },
43
+ "registered_name": "keras_hub>Llama3Tokenizer"
44
+ },
45
+ "config_file": "preprocessor.json",
46
+ "sequence_length": 1024,
47
+ "add_start_token": true,
48
+ "add_end_token": true
49
+ },
50
+ "registered_name": "keras_hub>Llama3CausalLMPreprocessor"
51
+ }
task.json ADDED
@@ -0,0 +1,82 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "module": "keras_hub.src.models.llama3.llama3_causal_lm",
3
+ "class_name": "Llama3CausalLM",
4
+ "config": {
5
+ "backbone": {
6
+ "module": "keras_hub.src.models.llama3.llama3_backbone",
7
+ "class_name": "Llama3Backbone",
8
+ "config": {
9
+ "name": "llama3_backbone",
10
+ "trainable": true,
11
+ "vocabulary_size": 128256,
12
+ "num_layers": 16,
13
+ "num_query_heads": 32,
14
+ "hidden_dim": 2048,
15
+ "intermediate_dim": 8192,
16
+ "rope_max_wavelength": 500000.0,
17
+ "rope_position_scaling_factor": 1,
18
+ "rope_frequency_adjustment_factor": 32,
19
+ "rope_low_freq_factor": 1,
20
+ "rope_high_freq_factor": 4,
21
+ "rope_pretraining_sequence_length": 8192,
22
+ "num_key_value_heads": 8,
23
+ "layer_norm_epsilon": 1e-05,
24
+ "dropout": 0
25
+ },
26
+ "registered_name": "keras_hub>Llama3Backbone"
27
+ },
28
+ "preprocessor": {
29
+ "module": "keras_hub.src.models.llama3.llama3_causal_lm_preprocessor",
30
+ "class_name": "Llama3CausalLMPreprocessor",
31
+ "config": {
32
+ "name": "llama3_causal_lm_preprocessor",
33
+ "trainable": true,
34
+ "dtype": {
35
+ "module": "keras",
36
+ "class_name": "DTypePolicy",
37
+ "config": {
38
+ "name": "float32"
39
+ },
40
+ "registered_name": null
41
+ },
42
+ "tokenizer": {
43
+ "module": "keras_hub.src.models.llama3.llama3_tokenizer",
44
+ "class_name": "Llama3Tokenizer",
45
+ "config": {
46
+ "name": "llama3_tokenizer_1",
47
+ "trainable": true,
48
+ "dtype": {
49
+ "module": "keras",
50
+ "class_name": "DTypePolicy",
51
+ "config": {
52
+ "name": "int32"
53
+ },
54
+ "registered_name": null
55
+ },
56
+ "config_file": "tokenizer.json",
57
+ "sequence_length": null,
58
+ "add_prefix_space": false,
59
+ "unsplittable_tokens": [
60
+ "<|finetune_right_pad_id|>",
61
+ "<|end_header_id|>",
62
+ "<|end_of_text|>",
63
+ "<|begin_of_text|>",
64
+ "<|eom_id|>",
65
+ "<|start_header_id|>",
66
+ "<|eot_id|>",
67
+ "<|python_tag|>"
68
+ ]
69
+ },
70
+ "registered_name": "keras_hub>Llama3Tokenizer"
71
+ },
72
+ "config_file": "preprocessor.json",
73
+ "sequence_length": 1024,
74
+ "add_start_token": true,
75
+ "add_end_token": true
76
+ },
77
+ "registered_name": "keras_hub>Llama3CausalLMPreprocessor"
78
+ },
79
+ "name": "llama3_causal_lm"
80
+ },
81
+ "registered_name": "keras_hub>Llama3CausalLM"
82
+ }
tokenizer.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "module": "keras_hub.src.models.llama3.llama3_tokenizer",
3
+ "class_name": "Llama3Tokenizer",
4
+ "config": {
5
+ "name": "llama3_tokenizer_1",
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
+ "<|begin_of_text|>",
20
+ "<|finetune_right_pad_id|>",
21
+ "<|start_header_id|>",
22
+ "<|end_of_text|>",
23
+ "<|python_tag|>",
24
+ "<|eot_id|>",
25
+ "<|eom_id|>",
26
+ "<|end_header_id|>"
27
+ ]
28
+ },
29
+ "registered_name": "keras_hub>Llama3Tokenizer"
30
+ }