SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

fix lm lables in docstring (#3529)

Browse Source
This commit is contained in:
Patrick von Platen
2020-03-30 14:26:24 +02:00
committed by GitHub
parent 75ec6c9e3a
commit 296252c49e
2 changed files with 5 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
src/transformers/modeling_t5.py
View File

@@ -900,8 +900,10 @@ class T5ForConditionalGeneration(T5PreTrainedModel):
r""" r"""
lm_labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`, defaults to :obj:`None`): lm_labels (:obj:`torch.LongTensor` of shape :obj:`(batch_size,)`, `optional`, defaults to :obj:`None`):
Labels for computing the sequence classification/regression loss. Labels for computing the sequence classification/regression loss.
Indices should be in :obj:`[0, ..., config.vocab_size - 1]`. Indices should be in :obj:`[-100, 0, ..., config.vocab_size - 1]`.
If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy). If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
All labels set to ``-100`` are ignored (masked), the loss is only
computed for labels in ``[0, ..., config.vocab_size]``
Returns: Returns:
:obj:`tuple(torch.FloatTensor)` comprising various elements depending on the configuration (:class:`~transformers.T5Config`) and inputs. :obj:`tuple(torch.FloatTensor)` comprising various elements depending on the configuration (:class:`~transformers.T5Config`) and inputs.

9
src/transformers/modeling_tf_t5.py
View File

@@ -799,11 +799,6 @@ class TFT5ForConditionalGeneration(TFT5PreTrainedModel):
@add_start_docstrings_to_callable(T5_INPUTS_DOCSTRING) @add_start_docstrings_to_callable(T5_INPUTS_DOCSTRING)
def call(self, decoder_input_ids, **kwargs): def call(self, decoder_input_ids, **kwargs):
r""" r"""
lm_labels (:obj:`tf.Tensor` of shape :obj:`(batch_size,)`, `optional`, defaults to :obj:`None`):
Labels for computing the sequence classification/regression loss.
Indices should be in :obj:`[0, ..., config.vocab_size - 1]`.
If :obj:`config.num_labels > 1` a classification loss is computed (Cross-Entropy).
Return: Return:
:obj:`tuple(tf.Tensor)` comprising various elements depending on the configuration (:class:`~transformers.T5Config`) and inputs. :obj:`tuple(tf.Tensor)` comprising various elements depending on the configuration (:class:`~transformers.T5Config`) and inputs.
loss (:obj:`tf.Tensor` of shape :obj:`(1,)`, `optional`, returned when :obj:`lm_label` is provided): loss (:obj:`tf.Tensor` of shape :obj:`(1,)`, `optional`, returned when :obj:`lm_label` is provided):
@@ -828,8 +823,8 @@ class TFT5ForConditionalGeneration(TFT5PreTrainedModel):
tokenizer = T5Tokenizer.from_pretrained('t5-small') tokenizer = T5Tokenizer.from_pretrained('t5-small')
model = TFT5ForConditionalGeneration.from_pretrained('t5-small') model = TFT5ForConditionalGeneration.from_pretrained('t5-small')
input_ids = tokenizer.encode("Hello, my dog is cute", return_tensors="tf") # Batch size 1 input_ids = tokenizer.encode("Hello, my dog is cute", return_tensors="tf") # Batch size 1
outputs = model(input_ids, input_ids=input_ids, lm_labels=input_ids) outputs = model(input_ids, input_ids=input_ids)
prediction_scores = outputs[:1] prediction_scores = outputs[0]
tokenizer = T5Tokenizer.from_pretrained('t5-small') tokenizer = T5Tokenizer.from_pretrained('t5-small')
model = TFT5ForConditionalGeneration.from_pretrained('t5-small') model = TFT5ForConditionalGeneration.from_pretrained('t5-small')