def train_dataloader(self) -> DataLoader:
dataloader = self.get_dataloader("train", batch_size=self.hparams.train_batch_size)
t_total = (
(len(dataloader.dataset) // (self.hparams.train_batch_size * max(1, self.hparams.n_gpu)))
// self.hparams.gradient_accumulation_steps
* float(self.hparams.num_train_epochs)
)
scheduler = get_linear_schedule_with_warmup(
self.opt, num_warmup_steps=self.hparams.warmup_steps, num_training_steps=t_total
)
self.lr_scheduler = scheduler
return dataloader
def train_dataloader(self):
train_dataset = SummarizationDataset(
self.tokenizer, data_dir=self.hparams.data_dir, type_path="train", block_size=self.hparams.max_seq_length
)
dataloader = DataLoader(train_dataset, batch_size=self.hparams.train_batch_size, num_workers=2, pin_memory=True, shuffle=True)
t_total = (
(len(dataloader.dataset) // (self.hparams.train_batch_size * max(1, self.hparams.n_gpu)))
// self.hparams.gradient_accumulation_steps
* float(self.hparams.num_train_epochs)
)
scheduler = get_linear_schedule_with_warmup(
self.opt, num_warmup_steps=self.hparams.warmup_steps, num_training_steps=t_total
)
self.lr_scheduler = scheduler
return dataloader
def train_dataloader(self):
train_dataset = MaskedPlotFactEmbeddingDatasetBart(
self.tokenizer,
args,
self.hparams.data_dir + "train/",
block_size=self.hparams.max_seq_length,
is_train=False)
dataloader = DataLoader(train_dataset,
batch_size=self.hparams.train_batch_size)
t_total = (
(len(dataloader.dataset) //
(self.hparams.train_batch_size * max(1, self.hparams.n_gpu))) //
self.hparams.gradient_accumulation_steps *
float(self.hparams.num_train_epochs))
scheduler = get_linear_schedule_with_warmup(
self.opt,
num_warmup_steps=self.hparams.warmup_steps,
num_training_steps=t_total)
self.lr_scheduler = scheduler
return dataloader