Upload transformerdecoder.py

transformerdecoder.py

@@ -47,6 +47,7 @@ class AugmentedPositionGPTConfig(PretrainedConfig):
         d_model=128,
         num_heads=2,
         num_layers=1,
+        num_hidden_layers = 128*4,
         max_position_embeddings=512,
         **kwargs,
     ):
@@ -188,7 +189,8 @@ class AugmentedPositionGPTForCausalLM(PreTrainedModel, GenerationMixin):
     
     
 
-    def forward(self, input_ids=None, attention_mask=None, input_embeds=None, labels=None, output_hidden_states=False, return_dict=True):
+    def forward(self, input_ids=None, attention_mask=None, input_embeds=None, labels=None, output_hidden_states=False, return_dict=True, cache_position=None, position_ids=None):
+        cache_position=None
         outputs= self.transformerdecoder(
             input_ids=input_ids,
             attention_mask=attention_mask,