Create configuration_ltgbert.py
Browse files- configuration_ltgbert.py +36 -0
configuration_ltgbert.py
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from transformers.configuration_utils import PretrainedConfig
|
| 2 |
+
|
| 3 |
+
|
| 4 |
+
class LtgbertConfig(PretrainedConfig):
|
| 5 |
+
"""Configuration class to store the configuration of a `LtgbertModel`.
|
| 6 |
+
"""
|
| 7 |
+
def __init__(
|
| 8 |
+
self,
|
| 9 |
+
vocab_size=32768,
|
| 10 |
+
attention_probs_dropout_prob=0.1,
|
| 11 |
+
hidden_dropout_prob=0.1,
|
| 12 |
+
hidden_size=768,
|
| 13 |
+
intermediate_size=2048,
|
| 14 |
+
max_position_embeddings=512,
|
| 15 |
+
position_bucket_size=32,
|
| 16 |
+
num_attention_heads=12,
|
| 17 |
+
num_hidden_layers=12,
|
| 18 |
+
layer_norm_eps=1.0e-7,
|
| 19 |
+
output_all_encoded_layers=True,
|
| 20 |
+
temperature=1.0,
|
| 21 |
+
**kwargs,
|
| 22 |
+
):
|
| 23 |
+
super().__init__(**kwargs)
|
| 24 |
+
|
| 25 |
+
self.vocab_size = vocab_size
|
| 26 |
+
self.hidden_size = hidden_size
|
| 27 |
+
self.num_hidden_layers = num_hidden_layers
|
| 28 |
+
self.num_attention_heads = num_attention_heads
|
| 29 |
+
self.intermediate_size = intermediate_size
|
| 30 |
+
self.hidden_dropout_prob = hidden_dropout_prob
|
| 31 |
+
self.attention_probs_dropout_prob = attention_probs_dropout_prob
|
| 32 |
+
self.max_position_embeddings = max_position_embeddings
|
| 33 |
+
self.output_all_encoded_layers = output_all_encoded_layers
|
| 34 |
+
self.position_bucket_size = position_bucket_size
|
| 35 |
+
self.layer_norm_eps = layer_norm_eps
|
| 36 |
+
self.temperature = temperature
|