Create rita_configuration.py
Browse files- rita_configuration.py +29 -0
rita_configuration.py
ADDED
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from transformers.configuration_utils import PretrainedConfig
|
2 |
+
from transformers.utils import logging
|
3 |
+
|
4 |
+
logger = logging.get_logger(__name__)
|
5 |
+
|
6 |
+
class RITAConfig(PretrainedConfig):
|
7 |
+
model_type = "rita"
|
8 |
+
|
9 |
+
def __init__(
|
10 |
+
self,
|
11 |
+
vocab_size=26,
|
12 |
+
d_model=1536,
|
13 |
+
num_layers=24,
|
14 |
+
max_seq_len=1024,
|
15 |
+
num_heads=24,
|
16 |
+
dropout=0.,
|
17 |
+
ff_ratio=4,
|
18 |
+
eos_token_id=2,
|
19 |
+
**kwargs,
|
20 |
+
):
|
21 |
+
super().__init__(eos_token_id=eos_token_id, **kwargs)
|
22 |
+
self.vocab_size = vocab_size
|
23 |
+
self.d_model = d_model
|
24 |
+
self.num_heads = num_heads
|
25 |
+
self.d_feedforward = d_model*ff_ratio
|
26 |
+
self.num_layers = num_layers
|
27 |
+
self.max_seq_len=max_seq_len
|
28 |
+
self.dropout = dropout
|
29 |
+
self.eos_token_id=eos_token_id
|