dnrops
/
transformers
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge pull request #2068 from huggingface/fix-2042 

Nicer error message when Bert's input is missing batch size
This commit is contained in:
Thomas Wolf 2019-12-06 12:06:42 +01:00 committed by GitHub
parent 35ff345fc9 18fb93530b
commit fc1bb1f867
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 10 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
transformers/modeling_bert.py
View File

@ -667,11 +667,10 @@ class BertModel(BertPreTrainedModel):
# ourselves in which case we just need to make it broadcastable to all heads.
if attention_mask.dim() == 3:
extended_attention_mask = attention_mask[:, None, :, :]
elif attention_mask.dim() == 2:
# Provided a padding mask of dimensions [batch_size, seq_length]
# - if the model is a decoder, apply a causal mask in addition to the padding mask
# - if the model is an encoder, make the mask broadcastable to [batch_size, num_heads, seq_length, seq_length]
if attention_mask.dim() == 2:
if self.config.is_decoder:
batch_size, seq_length = input_shape
seq_ids = torch.arange(seq_length, device=device)
@ -679,6 +678,8 @@ class BertModel(BertPreTrainedModel):
extended_attention_mask = causal_mask[:, None, :, :] * attention_mask[:, None, None, :]
else:
extended_attention_mask = attention_mask[:, None, None, :]
else:
raise ValueError("Wrong shape for input_ids (shape {}) or attention_mask (shape {})".format(input_shape, attention_mask.shape))
# Since attention_mask is 1.0 for positions we want to attend and 0.0 for
# masked positions, this operation will create a tensor which is 0.0 for
@ -696,8 +697,11 @@ class BertModel(BertPreTrainedModel):
if encoder_attention_mask.dim() == 3:
encoder_extended_attention_mask = encoder_attention_mask[:, None, :, :]
if encoder_attention_mask.dim() == 2:
elif encoder_attention_mask.dim() == 2:
encoder_extended_attention_mask = encoder_attention_mask[:, None, None, :]
else:
raise ValueError("Wrong shape for input_ids (shape {}) or encoder_attention_mask (shape {})".format(input_shape,
encoder_attention_mask.shape))
encoder_extended_attention_mask = encoder_extended_attention_mask.to(dtype=next(self.parameters()).dtype) # fp16 compatibility
encoder_extended_attention_mask = (1.0 - encoder_extended_attention_mask) * -10000.0