Upload configuration_mrt5.py with huggingface_hub
Browse files- configuration_mrt5.py +32 -0
configuration_mrt5.py
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from transformers.models.t5.configuration_t5 import T5Config
|
| 2 |
+
|
| 3 |
+
class MrT5Config(T5Config):
|
| 4 |
+
model_type = "mrt5"
|
| 5 |
+
def __init__(
|
| 6 |
+
self,
|
| 7 |
+
*args,
|
| 8 |
+
sigmoid_mask_scale=-10.0,
|
| 9 |
+
gate_layer_norm=True,
|
| 10 |
+
deletion_threshold=None,
|
| 11 |
+
delete_gate_layer=2,
|
| 12 |
+
use_softmax1=False,
|
| 13 |
+
deletion_type=None,
|
| 14 |
+
random_deletion_probability=0.5,
|
| 15 |
+
fixed_deletion_amount=0.5,
|
| 16 |
+
train_language="en",
|
| 17 |
+
eval_language="en",
|
| 18 |
+
use_gumbel_noise=False,
|
| 19 |
+
**kwargs,
|
| 20 |
+
):
|
| 21 |
+
super().__init__(*args, **kwargs)
|
| 22 |
+
self.deletion_threshold = deletion_threshold
|
| 23 |
+
self.sigmoid_mask_scale = sigmoid_mask_scale
|
| 24 |
+
self.gate_layer_norm = gate_layer_norm
|
| 25 |
+
self.use_softmax1 = use_softmax1
|
| 26 |
+
self.deletion_type = deletion_type
|
| 27 |
+
self.random_deletion_probability = random_deletion_probability
|
| 28 |
+
self.fixed_deletion_amount = fixed_deletion_amount
|
| 29 |
+
self.train_language = train_language
|
| 30 |
+
self.eval_language = eval_language
|
| 31 |
+
self.delete_gate_layer = delete_gate_layer
|
| 32 |
+
self.use_gumbel_noise = use_gumbel_noise
|