gugarosa commited on
Commit
b6a7e2f
1 Parent(s): 8ab0f29

Upload modeling_mixformer_sequential.py

Browse files
Files changed (1) hide show
  1. modeling_mixformer_sequential.py +1 -1
modeling_mixformer_sequential.py CHANGED
@@ -753,7 +753,7 @@ class MixFormerSequentialForCausalLM(MixFormerSequentialPreTrainedModel):
753
  **kwargs,
754
  ) -> CausalLMOutputWithPast:
755
  if attention_mask is not None and self.training:
756
- raise ValueError("`attention_mask` is not supported during training.")
757
 
758
  if past_key_values is None and attention_mask is None:
759
  lm_logits = self.layers(input_ids)
 
753
  **kwargs,
754
  ) -> CausalLMOutputWithPast:
755
  if attention_mask is not None and self.training:
756
+ print("`attention_mask` is not supported during training. Using it might lead to unexpected results.")
757
 
758
  if past_key_values is None and attention_mask is None:
759
  lm_logits = self.layers(input_ids)