Upload EncT5ForSequenceClassification
Browse files- model.safetensors +1 -1
- modeling_enct5.py +7 -2
model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 476301088
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e67a80a5bd78ab3885be58d157623db44c1ff78204817af457bd31b00e6b49aa
|
3 |
size 476301088
|
modeling_enct5.py
CHANGED
@@ -93,6 +93,7 @@ class EncT5ForSequenceClassification(EncT5PreTrainedModel):
|
|
93 |
|
94 |
# Initiate decoder embedding from scratch and define the corresponding latent vector vocabulary size.
|
95 |
self.decoder_embeddings = nn.Embedding(config.decoder_vocab_size, config.d_model)
|
|
|
96 |
|
97 |
# Initiate decoder projection head from scratch.
|
98 |
if config.problem_type == "multi_label_classification":
|
@@ -107,14 +108,18 @@ class EncT5ForSequenceClassification(EncT5PreTrainedModel):
|
|
107 |
|
108 |
def load_weights_from_pretrained_t5(self, model_path: str):
|
109 |
pretrained_t5_model = T5Model.from_pretrained(model_path)
|
110 |
-
|
|
|
|
|
|
|
|
|
|
|
111 |
|
112 |
def prepare_for_fine_tuning(self):
|
113 |
r"""
|
114 |
Prepares the model for fine-tuning by re-initializing the necessary weights for fine-tuning. This step should be
|
115 |
performed after loading the pre-trained T5 model but before fine-tuning.
|
116 |
"""
|
117 |
-
self.transformer.get_decoder().set_input_embeddings(self.decoder_embeddings)
|
118 |
self.transformer.get_decoder().apply(self._init_weights)
|
119 |
self._init_weights(self.classification_head)
|
120 |
|
|
|
93 |
|
94 |
# Initiate decoder embedding from scratch and define the corresponding latent vector vocabulary size.
|
95 |
self.decoder_embeddings = nn.Embedding(config.decoder_vocab_size, config.d_model)
|
96 |
+
self.transformer.get_decoder().set_input_embeddings(self.decoder_embeddings)
|
97 |
|
98 |
# Initiate decoder projection head from scratch.
|
99 |
if config.problem_type == "multi_label_classification":
|
|
|
108 |
|
109 |
def load_weights_from_pretrained_t5(self, model_path: str):
|
110 |
pretrained_t5_model = T5Model.from_pretrained(model_path)
|
111 |
+
|
112 |
+
# Override the decoder embedding weights to make them the correct shape.
|
113 |
+
pretrained_state_dict = pretrained_t5_model.state_dict()
|
114 |
+
pretrained_state_dict["decoder.embed_tokens.weight"] = self.decoder_embeddings.state_dict()["weight"]
|
115 |
+
|
116 |
+
self.transformer.load_state_dict(pretrained_state_dict, strict=False)
|
117 |
|
118 |
def prepare_for_fine_tuning(self):
|
119 |
r"""
|
120 |
Prepares the model for fine-tuning by re-initializing the necessary weights for fine-tuning. This step should be
|
121 |
performed after loading the pre-trained T5 model but before fine-tuning.
|
122 |
"""
|
|
|
123 |
self.transformer.get_decoder().apply(self._init_weights)
|
124 |
self._init_weights(self.classification_head)
|
125 |
|