def fill_mask(self, masked_input: str, topk: int = 5):
masked_token = '<mask>'
assert masked_token in masked_input and masked_input.count(masked_token) == 1, \
"Please add one {0} token for the input, eg: 'He is a {0} guy'".format(masked_token)
tokens = self.task.source_dictionary.encode_line(
'<s> ' + masked_input,
append_eos=True,
)
masked_index = (tokens == self.task.mask_idx).nonzero()
if tokens.dim() == 1:
tokens = tokens.unsqueeze(0)
with utils.eval(self.model):
features, extra = self.model(
tokens.long().to(device=self.device),
features_only=False,
return_all_hiddens=False,
)
logits = features[0, masked_index, :].squeeze()
prob = logits.softmax(dim=0)
values, index = prob.topk(k=topk, dim=0)
topk_predicted_token = self.task.source_dictionary.string(index)
topk_filled_outputs = []
for index, predicted_token in enumerate(
topk_predicted_token.split(' ')):
topk_filled_outputs.append((
masked_input.replace(masked_token, predicted_token),
values[index].item(),
predicted_token,
))
return topk_filled_outputs
def fill_mask(self, masked_input: str, topk: int = 5):
masked_token = "<mask>"
assert (
masked_token in masked_input
and masked_input.count(masked_token) == 1
), "Please add one {0} token for the input, eg: 'He is a {0} guy'".format(
masked_token)
text_spans = masked_input.split(masked_token)
text_spans_bpe = ((" {0} ".format(masked_token)).join([
self.bpe.encode(text_span.rstrip()) for text_span in text_spans
]).strip())
tokens = self.task.source_dictionary.encode_line(
"<s> " + text_spans_bpe + " </s>",
append_eos=False,
add_if_not_exist=False,
)
masked_index = (tokens == self.task.mask_idx).nonzero()
if tokens.dim() == 1:
tokens = tokens.unsqueeze(0)
with utils.eval(self.model):
features, extra = self.model(
tokens.long().to(device=self.device),
features_only=False,
return_all_hiddens=False,
)
logits = features[0, masked_index, :].squeeze()
prob = logits.softmax(dim=0)
values, index = prob.topk(k=topk, dim=0)
topk_predicted_token_bpe = self.task.source_dictionary.string(index)
topk_filled_outputs = []
for index, predicted_token_bpe in enumerate(
topk_predicted_token_bpe.split(" ")):
predicted_token = self.bpe.decode(predicted_token_bpe)
# Quick hack to fix https://github.com/pytorch/fairseq/issues/1306
if predicted_token_bpe.startswith("\u2581"):
predicted_token = " " + predicted_token
if " {0}".format(masked_token) in masked_input:
topk_filled_outputs.append((
masked_input.replace(" {0}".format(masked_token),
predicted_token),
values[index].item(),
predicted_token,
))
else:
topk_filled_outputs.append((
masked_input.replace(masked_token, predicted_token),
values[index].item(),
predicted_token,
))
return topk_filled_outputs
def disambiguate_pronoun(self, sentence: str) -> bool:
"""
Usage::
>>> disambiguate_pronoun('The _trophy_ would not fit in the brown suitcase because [it] was too big.')
True
>>> disambiguate_pronoun('The trophy would not fit in the brown suitcase because [it] was too big.')
'The trophy'
"""
assert hasattr(self.task, 'disambiguate_pronoun'), \
'roberta.disambiguate_pronoun() requires a model trained with the WSC task.'
with utils.eval(self.model):
return self.task.disambiguate_pronoun(self.model, sentence, use_cuda=self.device.type == 'cuda')
def _get_loss(self, sample, model, criterion):
assert hasattr(criterion, 'compute_loss'), \
'language_model_moe task requires the criterion to implement the compute_loss() method'
bsz = sample['target'].size(0)
src_tokens = sample['net_input']['src_tokens']
src_lengths = sample['net_input']['src_lengths']
#### E-STEP
with utils.eval(model): # disable dropout
with torch.no_grad(): # disable autograd
net_output = model(src_tokens=src_tokens,
src_lengths=src_lengths)
# pass net output to gating network to compute expert probabilities
expert_probs = model.gating_network(net_output)
# hard selection of experts
expert_assignments = [
self.expert_index(x) for x in expert_probs.argmax(dim=1)
]
# add expert assignments as BOS tokens
src_tokens[:, 0] = torch.Tensor(expert_assignments).long()
#### M-STEP
net_output = model(src_tokens=src_tokens, src_lengths=src_lengths)
loss, _ = criterion.compute_loss(model,
net_output,
sample,
reduce=False)
loss = loss.view(sample['target'].size(0), -1)
loss = loss.sum(dim=1, keepdim=True)
loss = loss.sum()
sample_size = sample['target'].size(
0) if self.args.sentence_avg else sample['ntokens']
logging_output = {
'loss': utils.item(loss.data),
'ntokens': sample['ntokens'],
'nsentences': bsz,
'sample_size': sample_size,
"expert_assignments": expert_probs.argmax(dim=1)
}
return loss, sample_size, logging_output
def fill_single_mask(self, masked_inputs, topk=3):
if isinstance(masked_inputs, str):
masked_inputs = [masked_inputs]
assert all(self.masked_token in masked_input for masked_input in masked_inputs), \
"Please add one {0} token for the input, eg: 'He is a {0} guy'".format(self.masked_token)
tokens = [
self.encode_masked_input(masked_input)
for masked_input in masked_inputs
]
pad_to_length = max(len(token) for token in tokens)
tokens = data_utils.collate_tokens(
tokens,
self.task.source_dictionary.pad(),
self.task.source_dictionary.eos(),
False,
False,
pad_to_length=pad_to_length,
)
if tokens.dim() == 1:
tokens = tokens.unsqueeze(0)
src_lengths = tokens.ne(self.task.source_dictionary.pad()).sum(dim=-1)
masked_tokens = tokens.eq(self.task.source_dictionary.mask_index)
# with utils.model_eval(self.model): # new version
with utils.eval(self.model):
logits = self.model.forward_encoder(
tokens.long().to(device=self.device),
src_lengths=src_lengths.to(device=self.device),
masked_tokens=masked_tokens)
prob = logits.softmax(dim=-1)
all_values, all_index = prob.topk(k=topk, dim=-1)
topk_predicted_token_bpe = self.task.source_dictionary.string(
all_index)
topk_predicted_token_bpe = [
tokens.split(' ')
for tokens in topk_predicted_token_bpe.split('\n')
]
return topk_predicted_token_bpe
def _get_loss(self, sample, model, criterion):
assert hasattr(criterion, 'compute_loss'), \
'translation_moe task requires the criterion to implement the compute_loss() method'
k = self.args.num_experts
bsz = sample['target'].size(0)
def get_lprob_y(encoder_out, prev_output_tokens_k):
net_output = model.decoder(
prev_output_tokens=prev_output_tokens_k,
encoder_out=encoder_out,
)
loss, _ = criterion.compute_loss(model,
net_output,
sample,
reduce=False)
loss = loss.view(bsz, -1)
return -loss.sum(dim=1, keepdim=True) # -> B x 1
def get_lprob_yz(winners=None):
encoder_out = model.encoder(
src_tokens=sample['net_input']['src_tokens'],
src_lengths=sample['net_input']['src_lengths'],
)
if winners is None:
lprob_y = []
for i in range(k):
prev_output_tokens_k = sample['net_input'][
'prev_output_tokens'].clone()
assert not prev_output_tokens_k.requires_grad
prev_output_tokens_k[:, 0] = self.expert_index(i)
lprob_y.append(
get_lprob_y(encoder_out, prev_output_tokens_k))
lprob_y = torch.cat(lprob_y, dim=1) # -> B x K
else:
prev_output_tokens_k = sample['net_input'][
'prev_output_tokens'].clone()
prev_output_tokens_k[:, 0] = self.expert_index(winners)
lprob_y = get_lprob_y(encoder_out,
prev_output_tokens_k) # -> B
if self.uniform_prior:
lprob_yz = lprob_y
else:
lprob_z = model.gating_network(encoder_out) # B x K
if winners is not None:
lprob_z = lprob_z.gather(dim=1,
index=winners.unsqueeze(-1))
lprob_yz = lprob_y + lprob_z.type_as(lprob_y) # B x K
return lprob_yz
# compute responsibilities without dropout
with utils.eval(model): # disable dropout
with torch.no_grad(): # disable autograd
lprob_yz = get_lprob_yz() # B x K
prob_z_xy = torch.nn.functional.softmax(lprob_yz, dim=1)
assert not prob_z_xy.requires_grad
# compute loss with dropout
if self.hard_selection:
winners = prob_z_xy.max(dim=1)[1]
loss = -get_lprob_yz(winners)
else:
lprob_yz = get_lprob_yz() # B x K
loss = -LogSumExpMoE.apply(lprob_yz, prob_z_xy, 1)
loss = loss.sum()
sample_size = sample['target'].size(
0) if self.args.sentence_avg else sample['ntokens']
logging_output = {
'loss': utils.item(loss.data),
'ntokens': sample['ntokens'],
'nsentences': bsz,
'sample_size': sample_size,
'posterior': prob_z_xy.float().sum(dim=0).cpu(),
}
return loss, sample_size, logging_output
def fill_noised_mask(self, masked_inputs: List[str], topk=1):
masked_token = '<mask>'
noises, topk_opt = [], []
text_spans = [sent.split(masked_token) for src, sent in masked_inputs]
noised_tokens = []
targets_bpe = []
for (src, _), segs in zip(masked_inputs, text_spans):
bpe_src = self.bpe.encode(src.strip())
bpe_tgt = ' {0} '.format(masked_token).join(
[self.bpe.encode(seg.rstrip()) for seg in segs])
bpe_idx = self.task.source_dictionary.encode_line(
'<s> ' + bpe_src + ' </s> </s> ' + bpe_tgt + ' </s>',
append_eos=False,
add_if_not_exist=False,
)
tgt_bpe_idx = self.task.source_dictionary.encode_line(
'<s> ' + bpe_tgt + ' </s>',
append_eos=False,
add_if_not_exist=False,
)
noised_tokens.append(bpe_idx)
targets_bpe.append(tgt_bpe_idx)
sample = self._build_sample(noised_tokens).long()
masked_index = (sample == self.task.mask_idx)
with utils.eval(self.model):
# features: B x T x |V|
features, extra = self.model(sample,
features_only=False,
return_all_hiddens=False,
masked_tokens=masked_index)
prob = features.softmax(dim=-1)
# values, index = prob.topk(k=topk, dim=-1)
values, index = prob.max(dim=-1)
index = index.squeeze(-1) # K
extra_symbols_to_ignore = set([])
extra_symbols_to_ignore.add(
self.task.source_dictionary[self.task.source_dictionary.eos()])
extra_symbols_to_ignore.add(
self.task.source_dictionary[self.task.source_dictionary.bos()])
tot_masks = 0
for ii, sent in enumerate(targets_bpe):
decode_noise_tokens = self.decode(sent)
decode_noise_tokens = decode_noise_tokens.replace(
"<mask>", " <mask>").strip()
K = masked_index[ii, :].sum().item()
topk_predictions = index[tot_masks:tot_masks + K]
tot_masks += K
assert len(topk_predictions) == decode_noise_tokens.split(
" ").count('<mask>')
output = []
mask_count = 0
topk_predicted_token_bpe = self.task.source_dictionary.string(
topk_predictions, skip_ignore=True).split()
for token in decode_noise_tokens.split(" "):
if token == "<mask>":
predict_bpe = topk_predicted_token_bpe[mask_count]
if predict_bpe in extra_symbols_to_ignore:
continue
predicted_token = self.bpe.decode(predict_bpe)
# output.append("[" + predicted_token.strip() + "]")
output.append(predicted_token.strip())
mask_count += 1
else:
output.append(token.strip())
topk_opt.append(" ".join(output))
noises.append(decode_noise_tokens)
return topk_opt, noises
def fill_mask(self, masked_inputs, topk=3, return_filled_sentence=False):
if isinstance(masked_inputs, str):
masked_inputs = [masked_inputs]
masked_token = '[MASK]'
assert all(masked_token in masked_input for masked_input in masked_inputs), \
"Please add one {0} token for the input, eg: 'He is a {0} guy'".format(masked_token)
def encode_masked_input(masked_input):
text_spans = masked_input.split(masked_token)
text_spans_bpe = (' {0} '.format(masked_token)).join([
self.bpe.encode(text_span.rstrip()) for text_span in text_spans
]).strip()
tokens = self.task.source_dictionary.encode_line(
'[CLS] ' + text_spans_bpe + ' [SEP]',
append_eos=False,
add_if_not_exist=False,
)
return tokens
tokens = [
encode_masked_input(masked_input) for masked_input in masked_inputs
]
pad_to_length = max(len(token) for token in tokens)
tokens = data_utils.collate_tokens(
tokens,
self.task.source_dictionary.pad(),
self.task.source_dictionary.eos(),
False,
False,
pad_to_length=pad_to_length,
)
if tokens.dim() == 1:
tokens = tokens.unsqueeze(0)
src_lengths = tokens.ne(self.task.source_dictionary.pad()).sum(dim=-1)
masked_tokens = tokens.eq(self.task.source_dictionary.mask_index)
# with utils.model_eval(self.model): # new version
with utils.eval(self.model):
logits = self.model.forward_encoder(
tokens.long().to(device=self.device),
src_lengths=src_lengths.to(device=self.device),
masked_tokens=masked_tokens)
prob = logits.softmax(dim=-1)
all_values, all_index = prob.topk(k=topk, dim=-1)
topk_predicted_token_bpe = self.task.source_dictionary.string(
all_index)
topk_predicted_token_bpe = [
tokens.split(' ')
for tokens in topk_predicted_token_bpe.split('\n')
]
if not return_filled_sentence:
return topk_predicted_token_bpe
# all_outputs = []
# topk_predicted_token_bpe = iter(topk_predicted_token_bpe)
# topk_filled_outputs = []
# for masked_input in masked_inputs:
# predicted_token = self.bpe.decode(predicted_token_bpe)
# if predicted_token_bpe.startswith('\u2581'):
# predicted_token = ' ' + predicted_token
# if " {0}".format(masked_token) in masked_input:
# topk_filled_outputs.append((
# masked_input.replace(
# ' {0}'.format(masked_token), predicted_token
# ),
# values[index].item(),
# predicted_token,
# ))
# else:
# topk_filled_outputs.append((
# masked_input.replace(masked_token, predicted_token),
# values[index].item(),
# predicted_token,
# ))
# all_outputs.append(topk_filled_outputs)
return None
