Fix @dataclass decorator bug reported at https://github.com/huggingface/transformers/issues/30412
#5
by
pamessina
- opened
- modeling_cxrbert.py +5 -2
modeling_cxrbert.py
CHANGED
|
@@ -14,11 +14,14 @@ from transformers.modeling_outputs import ModelOutput
|
|
| 14 |
|
| 15 |
from .configuration_cxrbert import CXRBertConfig
|
| 16 |
|
|
|
|
|
|
|
| 17 |
BERTTupleOutput = Tuple[T, T, T, T, T]
|
| 18 |
|
|
|
|
| 19 |
class CXRBertOutput(ModelOutput):
|
| 20 |
-
last_hidden_state: torch.FloatTensor
|
| 21 |
-
logits: torch.FloatTensor
|
| 22 |
cls_projected_embedding: Optional[torch.FloatTensor] = None
|
| 23 |
hidden_states: Optional[Tuple[torch.FloatTensor]] = None
|
| 24 |
attentions: Optional[Tuple[torch.FloatTensor]] = None
|
|
|
|
| 14 |
|
| 15 |
from .configuration_cxrbert import CXRBertConfig
|
| 16 |
|
| 17 |
+
from dataclasses import dataclass # manually added due to this bug: https://github.com/huggingface/transformers/issues/30412
|
| 18 |
+
|
| 19 |
BERTTupleOutput = Tuple[T, T, T, T, T]
|
| 20 |
|
| 21 |
+
@dataclass # manually added due to this bug: https://github.com/huggingface/transformers/issues/30412
|
| 22 |
class CXRBertOutput(ModelOutput):
|
| 23 |
+
last_hidden_state: torch.FloatTensor = None # None added. Not present in the original code
|
| 24 |
+
logits: torch.FloatTensor = None # None added. Not present in the original code
|
| 25 |
cls_projected_embedding: Optional[torch.FloatTensor] = None
|
| 26 |
hidden_states: Optional[Tuple[torch.FloatTensor]] = None
|
| 27 |
attentions: Optional[Tuple[torch.FloatTensor]] = None
|