Dejiao Z
commited on
Commit
·
a25a3cc
1
Parent(s):
e81a062
fixed config typo
Browse files- .ipynb_checkpoints/config-checkpoint.json +25 -0
- config.json +1 -1
.ipynb_checkpoints/config-checkpoint.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "codesage/codesage-large-v2",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"CodeSage"
|
| 5 |
+
],
|
| 6 |
+
"auto_map": {
|
| 7 |
+
"AutoConfig": "config_codesage.CodeSageConfig",
|
| 8 |
+
"AutoTokenizer": "tokenization_codesage.CodeSageTokenizer",
|
| 9 |
+
"AutoModel": "modeling_codesage.CodeSageModel",
|
| 10 |
+
"AutoModelForMaskedLM": "modeling_codesage.CodeSageForMaskedLM",
|
| 11 |
+
"AutoModelForSequenceClassification": "modeling_codesage.CodeSageForSequenceClassification"
|
| 12 |
+
},
|
| 13 |
+
"activation_function": "gelu_new",
|
| 14 |
+
"attention_dropout_prob": 0.1,
|
| 15 |
+
"embedding_dropout_prob": 0.1,
|
| 16 |
+
"initializer_range": 0.02,
|
| 17 |
+
"layer_norm_epsilon": 1e-05,
|
| 18 |
+
"hidden_size": 1024,
|
| 19 |
+
"num_attention_heads": 16,
|
| 20 |
+
"num_hidden_layers": 24,
|
| 21 |
+
"intermediate_size": 8192,
|
| 22 |
+
"max_position_embeddings": 2048,
|
| 23 |
+
"residual_dropout_prob": 0.1,
|
| 24 |
+
"vocab_size": 49154
|
| 25 |
+
}
|
config.json
CHANGED
|
@@ -15,7 +15,7 @@
|
|
| 15 |
"embedding_dropout_prob": 0.1,
|
| 16 |
"initializer_range": 0.02,
|
| 17 |
"layer_norm_epsilon": 1e-05,
|
| 18 |
-
"hidden_size":
|
| 19 |
"num_attention_heads": 16,
|
| 20 |
"num_hidden_layers": 24,
|
| 21 |
"intermediate_size": 8192,
|
|
|
|
| 15 |
"embedding_dropout_prob": 0.1,
|
| 16 |
"initializer_range": 0.02,
|
| 17 |
"layer_norm_epsilon": 1e-05,
|
| 18 |
+
"hidden_size": 2048,
|
| 19 |
"num_attention_heads": 16,
|
| 20 |
"num_hidden_layers": 24,
|
| 21 |
"intermediate_size": 8192,
|