Commit
·
b566f66
1
Parent(s):
b806a55
initial commit
Browse files- config.json +74 -0
- finetunedmodel.weights.h5 +3 -0
config.json
ADDED
|
@@ -0,0 +1,74 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"backbone": {
|
| 3 |
+
"module": "keras_hub.src.models.gemma.gemma_backbone",
|
| 4 |
+
"class_name": "GemmaBackbone",
|
| 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_hub>GemmaBackbone",
|
| 26 |
+
"build_config": {
|
| 27 |
+
"input_shape": null
|
| 28 |
+
}
|
| 29 |
+
},
|
| 30 |
+
"preprocessor": {
|
| 31 |
+
"module": "keras_hub.src.models.gemma.gemma_causal_lm_preprocessor",
|
| 32 |
+
"class_name": "GemmaCausalLMPreprocessor",
|
| 33 |
+
"config": {
|
| 34 |
+
"name": "gemma_causal_lm_preprocessor",
|
| 35 |
+
"trainable": true,
|
| 36 |
+
"dtype": {
|
| 37 |
+
"module": "keras",
|
| 38 |
+
"class_name": "DTypePolicy",
|
| 39 |
+
"config": {
|
| 40 |
+
"name": "float32"
|
| 41 |
+
},
|
| 42 |
+
"registered_name": null
|
| 43 |
+
},
|
| 44 |
+
"tokenizer": {
|
| 45 |
+
"module": "keras_hub.src.models.gemma.gemma_tokenizer",
|
| 46 |
+
"class_name": "GemmaTokenizer",
|
| 47 |
+
"config": {
|
| 48 |
+
"name": "gemma_tokenizer",
|
| 49 |
+
"trainable": true,
|
| 50 |
+
"dtype": {
|
| 51 |
+
"module": "keras",
|
| 52 |
+
"class_name": "DTypePolicy",
|
| 53 |
+
"config": {
|
| 54 |
+
"name": "int32"
|
| 55 |
+
},
|
| 56 |
+
"registered_name": null
|
| 57 |
+
},
|
| 58 |
+
"config_file": "tokenizer.json",
|
| 59 |
+
"proto": null,
|
| 60 |
+
"sequence_length": null,
|
| 61 |
+
"add_bos": false,
|
| 62 |
+
"add_eos": false
|
| 63 |
+
},
|
| 64 |
+
"registered_name": "keras_hub>GemmaTokenizer"
|
| 65 |
+
},
|
| 66 |
+
"config_file": "preprocessor.json",
|
| 67 |
+
"sequence_length": 512,
|
| 68 |
+
"add_start_token": true,
|
| 69 |
+
"add_end_token": true
|
| 70 |
+
},
|
| 71 |
+
"registered_name": "keras_hub>GemmaCausalLMPreprocessor"
|
| 72 |
+
},
|
| 73 |
+
"name": "gemma_causal_lm"
|
| 74 |
+
}
|
finetunedmodel.weights.h5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dd2db5cc122693c0ed4ce6b001661699b2b7fe995d56091f4c262d3fc16d8c5b
|
| 3 |
+
size 10036254200
|