def __iter__(self):
samplers_list = []
sampler_iterators = []
for dataset_idx in range(self.number_of_datasets):
cur_dataset = self.dataset.datasets[dataset_idx]
sampler = RandomSampler(cur_dataset)
samplers_list.append(sampler)
cur_sampler_iterator = sampler.__iter__()
sampler_iterators.append(cur_sampler_iterator)
push_index_val = [0] + self.dataset.cumulative_sizes[:-1]
step = self.batch_size * self.number_of_datasets
samples_to_grab = self.batch_size
epoch_samples = self.largest_dataset_size * self.number_of_datasets
final_samples_list = []
for _ in range(0, epoch_samples, step):
for i in range(self.number_of_datasets):
cur_batch_sampler = sampler_iterators[i]
cur_samples = []
for _ in range(samples_to_grab):
try:
cur_sample_org = cur_batch_sampler.__next__()
cur_sample = cur_sample_org + push_index_val[i]
cur_samples.append(cur_sample)
except StopIteration:
sampler_iterators[i] = sampelrs_list[i].__iter__()
cur_batch_sampler = sampler_iterators[i]
cur_sample_org = cur_batch_sampler.__next__()
cur_sample = cur_sample_org + push_index_val[i]
cur_samples.append(cur_sample)
final_samples_list.extend(cur_samples)
return iter(final_samples_list)
def __iter__(self):
samplers_list = []
sampler_iterators = []
datasets_length = []
for dataset_idx in range(self.number_of_datasets):
cur_dataset = self.dataset.datasets[dataset_idx]
if dataset_idx == 0:
# the first dataset is kept at RandomSampler
sampler = RandomSampler(cur_dataset)
else:
# the second unbalanced dataset is changed
sampler = ExampleImbalancedDatasetSampler(cur_dataset)
samplers_list.append(sampler)
cur_sampler_iterator = sampler.__iter__()
sampler_iterators.append(cur_sampler_iterator)
datasets_length.append(len(cur_dataset))
push_index_val = [0] + self.dataset.cumulative_sizes[:-1]
step = self.batch_size * self.number_of_datasets
samples_to_grab = self.batch_size
largest_dataset_index = torch.argmax(
torch.as_tensor(datasets_length)).item()
# for this case we want to get all samples in dataset, this force us to resample from the smaller datasets
epoch_samples = datasets_length[
largest_dataset_index] * self.number_of_datasets
final_samples_list = [
] # this is a list of indexes from the combined dataset
for _ in range(0, epoch_samples, step):
for i in range(self.number_of_datasets):
cur_batch_sampler = sampler_iterators[i]
cur_samples = []
for _ in range(samples_to_grab):
try:
cur_sample_org = cur_batch_sampler.__next__()
cur_sample = cur_sample_org + push_index_val[i]
cur_samples.append(cur_sample)
except StopIteration:
if i == largest_dataset_index:
# largest dataset iterator is done we can break
samples_to_grab = len(
cur_samples) # adjusting the samples_to_grab
break # got to the end of iterator - extend final list and continue to next task
else:
# restart the iterator - we want more samples until finishing with the largest dataset
sampler_iterators[i] = samplers_list[i].__iter__()
cur_batch_sampler = sampler_iterators[i]
cur_sample_org = cur_batch_sampler.__next__()
cur_sample = cur_sample_org + push_index_val[i]
cur_samples.append(cur_sample)
final_samples_list.extend(cur_samples)
return iter(final_samples_list)
def __iter__(self):
samplers_list = []
sampler_iterators = []
for dataset_idx in range(self.number_of_datasets):
cur_dataset = self.dataset.datasets[dataset_idx]
if dataset_idx == 0:
# the first dataset is kept at RandomSampler
sampler = RandomSampler(cur_dataset)
else:
# the second unbalanced dataset is changed
sampler = ExampleImbalancedDatasetSampler(cur_dataset)
samplers_list.append(sampler)
cur_sampler_iterator = sampler.__iter__()
sampler_iterators.append(cur_sampler_iterator)
push_index_val = [0] + self.dataset.cumulative_sizes[:-1]
step = self.batch_size * self.number_of_datasets
samples_to_grab = self.batch_size
# for this case we want to get all samples in dataset, this force us to resample from the smaller datasets
epoch_samples = self.largest_dataset_size * self.number_of_datasets
final_samples_list = [
] # this is a list of indexes from the combined dataset
for _ in range(0, epoch_samples, step):
for i in range(self.number_of_datasets):
cur_batch_sampler = sampler_iterators[i]
cur_samples = []
for _ in range(samples_to_grab):
try:
cur_sample_org = cur_batch_sampler.__next__()
cur_sample = cur_sample_org + push_index_val[i]
cur_samples.append(cur_sample)
except StopIteration:
# got to the end of iterator - restart the iterator and continue to get samples
# until reaching "epoch_samples"
sampler_iterators[i] = samplers_list[i].__iter__()
cur_batch_sampler = sampler_iterators[i]
cur_sample_org = cur_batch_sampler.__next__()
cur_sample = cur_sample_org + push_index_val[i]
cur_samples.append(cur_sample)
final_samples_list.extend(cur_samples)
return iter(final_samples_list)
def get_data_order(self, train_examples, data_order_seed,
num_train_epochs):
train_features = convert_examples_to_features(
train_examples,
self.label_map,
self.rparams.max_seq_length,
self.tokenizer,
verbose=False,
)
train_data, train_tokens = convert_to_dataset(
train_features,
label_mode=get_label_mode(self.label_map),
)
iterators = []
train_sampler = RandomSampler(train_data)
print("Setting data order seed to {}".format(data_order_seed))
torch.manual_seed(data_order_seed)
for epoch in range(num_train_epochs):
iterators.append(train_sampler.__iter__())
return iterators
def __init__(self, dataset: ConcatNamedDataset, batch_size: int):
self.dataset = dataset
self.batch_size = batch_size
self.number_of_datasets = dataset.get_datasets_nb()
self.samplers_list: List[Sampler] = []
self.sampler_iterators: List[Iterator] = []
self.datasets_length: List[int] = [] # as we get datasets in descending size, 0 is the largest dataset
self.datasets_by_desc_size = self.dataset.get_datasets_info_by_desc_count()
for idx, (lang_id, lang, dataset_count) in enumerate(self.datasets_by_desc_size):
cur_dataset = self.dataset.get_dataset_by_lang_id(lang_id)
logger.debug(
f"Sampling batches from Dataset[lang_id:{lang_id}, count:{dataset_count}, len:{len(cur_dataset)} lang:{lang}]"
)
sampler = RandomSampler(cur_dataset)
self.samplers_list.append(sampler)
cur_sampler_iterator = sampler.__iter__()
self.sampler_iterators.append(cur_sampler_iterator)
self.datasets_length.append(dataset_count)
self.generate_sample_list()
def __iter__(self):
for index in RandomSampler.__iter__(self):
yield self.data_source[index]
