User01110 commited on
Commit
d95a00a
·
verified ·
1 Parent(s): beb8d09

Accept tokenizer token_type_ids during generation

Browse files
Files changed (1) hide show
  1. modeling_tinyqwen3_novelty.py +7 -3
modeling_tinyqwen3_novelty.py CHANGED
@@ -141,20 +141,24 @@ class TinyQwen3NoveltyForCausalLM(PreTrainedModel, GenerationMixin):
141
  def set_input_embeddings(self, value):
142
  self.embed_tokens = value
143
 
144
- def prepare_inputs_for_generation(self, input_ids, attention_mask=None, **kwargs):
145
  max_len = getattr(self.config, "max_position_embeddings", None)
146
  if max_len is not None and input_ids.shape[-1] > max_len:
147
  input_ids = input_ids[:, -max_len:]
148
  if attention_mask is not None:
149
  attention_mask = attention_mask[:, -max_len:]
 
 
150
 
151
  model_inputs = {"input_ids": input_ids}
152
  if attention_mask is not None:
153
  model_inputs["attention_mask"] = attention_mask
 
 
154
  return model_inputs
155
 
156
- def forward(self, input_ids=None, attention_mask=None, labels=None, return_dict=True, **kwargs):
157
- del attention_mask, kwargs
158
  x = self.embed_tokens(input_ids)
159
  for layer in self.layers:
160
  x = layer(x)
 
141
  def set_input_embeddings(self, value):
142
  self.embed_tokens = value
143
 
144
+ def prepare_inputs_for_generation(self, input_ids, attention_mask=None, token_type_ids=None, **kwargs):
145
  max_len = getattr(self.config, "max_position_embeddings", None)
146
  if max_len is not None and input_ids.shape[-1] > max_len:
147
  input_ids = input_ids[:, -max_len:]
148
  if attention_mask is not None:
149
  attention_mask = attention_mask[:, -max_len:]
150
+ if token_type_ids is not None:
151
+ token_type_ids = token_type_ids[:, -max_len:]
152
 
153
  model_inputs = {"input_ids": input_ids}
154
  if attention_mask is not None:
155
  model_inputs["attention_mask"] = attention_mask
156
+ if token_type_ids is not None:
157
+ model_inputs["token_type_ids"] = token_type_ids
158
  return model_inputs
159
 
160
+ def forward(self, input_ids=None, attention_mask=None, token_type_ids=None, labels=None, return_dict=True, **kwargs):
161
+ del attention_mask, token_type_ids, kwargs
162
  x = self.embed_tokens(input_ids)
163
  for layer in self.layers:
164
  x = layer(x)