Hugging Face's logo

sultan-hassan
/
CosmoGemma_2b_en

Text Generation
KerasHub
gemma
Model card Files Community
1
sultan-hassan commited on
Commit
ff6b2e2
•
1 Parent(s): 0cdc8ca

Update config.json

Browse files
Files changed (1) hide show
  1. config.json +26 -24
config.json CHANGED
@@ -1,26 +1,28 @@
1
  {
2
- "module": "keras_nlp.src.models.gemma.gemma_backbone",
3
- "class_name": "GemmaBackbone",
4
- "model_type": "gemma",
5
- "config": {
6
- "name": "gemma_backbone",
7
- "trainable": true,
8
- "vocabulary_size": 256000,
9
- "num_layers": 18,
10
- "num_query_heads": 8,
11
- "num_key_value_heads": 1,
12
- "hidden_dim": 2048,
13
- "intermediate_dim": 32768,
14
- "head_dim": 256,
15
- "layer_norm_epsilon": 1e-06,
16
- "dropout": 0,
17
- "query_head_dim_normalize": true,
18
- "use_post_ffw_norm": false,
19
- "use_post_attention_norm": false,
20
- "final_logit_soft_cap": null,
21
- "attention_logit_soft_cap": null,
22
- "sliding_window_size": 4096,
23
- "use_sliding_window_attention": false
24
- },
25
- "registered_name": "keras_nlp>GemmaBackbone"
 
 
26
  }
 
1
  {
2
+ "architectures": [
3
+ "Gemma2ForCausalLM"
4
+ ],
5
+
6
+ "module": "keras_nlp.src.models.gemma.gemma_backbone",
7
+ "class_name": "GemmaBackbone",
8
+ "model_type": "gemma",
9
+ "name": "gemma_backbone",
10
+ "trainable": true,
11
+ "vocabulary_size": 256000,
12
+ "num_layers": 18,
13
+ "num_query_heads": 8,
14
+ "num_key_value_heads": 1,
15
+ "hidden_dim": 2048,
16
+ "intermediate_dim": 32768,
17
+ "head_dim": 256,
18
+ "layer_norm_epsilon": 1e-06,
19
+ "dropout": 0,
20
+ "query_head_dim_normalize": true,
21
+ "use_post_ffw_norm": false,
22
+ "use_post_attention_norm": false,
23
+ "final_logit_soft_cap": null,
24
+ "attention_logit_soft_cap": null,
25
+ "sliding_window_size": 4096,
26
+ "use_sliding_window_attention": false
27
+ "registered_name": "keras_nlp>GemmaBackbone"
28
  }