Unverified Commit 6f1727d8 authored by Gunjan Chhablani's avatar Gunjan Chhablani Committed by GitHub
Browse files

Fix Seq2SeqTrainingArguments docs (#16295)

* Indent Seq2Seq Train Args docs

* Add Args keyword to Seq2Seq Train Args docs
parent 7643b1ca
...@@ -27,20 +27,21 @@ logger = logging.getLogger(__name__) ...@@ -27,20 +27,21 @@ logger = logging.getLogger(__name__)
@add_start_docstrings(TrainingArguments.__doc__) @add_start_docstrings(TrainingArguments.__doc__)
class Seq2SeqTrainingArguments(TrainingArguments): class Seq2SeqTrainingArguments(TrainingArguments):
""" """
sortish_sampler (`bool`, *optional*, defaults to `False`): Args:
Whether to use a *sortish sampler* or not. Only possible if the underlying datasets are *Seq2SeqDataset* for sortish_sampler (`bool`, *optional*, defaults to `False`):
now but will become generally available in the near future. Whether to use a *sortish sampler* or not. Only possible if the underlying datasets are *Seq2SeqDataset*
for now but will become generally available in the near future.
It sorts the inputs according to lengths in order to minimize the padding size, with a bit of randomness for
the training set. It sorts the inputs according to lengths in order to minimize the padding size, with a bit of randomness
predict_with_generate (`bool`, *optional*, defaults to `False`): for the training set.
Whether to use generate to calculate generative metrics (ROUGE, BLEU). predict_with_generate (`bool`, *optional*, defaults to `False`):
generation_max_length (`int`, *optional*): Whether to use generate to calculate generative metrics (ROUGE, BLEU).
The `max_length` to use on each evaluation loop when `predict_with_generate=True`. Will default to the generation_max_length (`int`, *optional*):
`max_length` value of the model configuration. The `max_length` to use on each evaluation loop when `predict_with_generate=True`. Will default to the
generation_num_beams (`int`, *optional*): `max_length` value of the model configuration.
The `num_beams` to use on each evaluation loop when `predict_with_generate=True`. Will default to the generation_num_beams (`int`, *optional*):
`num_beams` value of the model configuration. The `num_beams` to use on each evaluation loop when `predict_with_generate=True`. Will default to the
`num_beams` value of the model configuration.
""" """
sortish_sampler: bool = field(default=False, metadata={"help": "Whether to use SortishSampler or not."}) sortish_sampler: bool = field(default=False, metadata={"help": "Whether to use SortishSampler or not."})
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment