fairseq/criterions/masked_lm.py

Survived

Mutant 2707

--- fairseq/criterions/masked_lm.py
+++ fairseq/criterions/masked_lm.py
@@ -12,7 +12,7 @@
 from fairseq.criterions import FairseqCriterion, register_criterion
 
 
-@register_criterion('masked_lm')
+@register_criterion('XXmasked_lmXX')
 class MaskedLmLoss(FairseqCriterion):
     """
     Implementation for the loss used in masked language model (MLM) training.

Mutant 2708

--- fairseq/criterions/masked_lm.py
+++ fairseq/criterions/masked_lm.py
@@ -11,8 +11,6 @@
 from fairseq import metrics, modules, utils
 from fairseq.criterions import FairseqCriterion, register_criterion
 
-
-@register_criterion('masked_lm')
 class MaskedLmLoss(FairseqCriterion):
     """
     Implementation for the loss used in masked language model (MLM) training.

Mutant 2709

--- fairseq/criterions/masked_lm.py
+++ fairseq/criterions/masked_lm.py
@@ -22,7 +22,7 @@
         super().__init__(task)
         self.tpu = tpu
 
-    def forward(self, model, sample, reduce=True):
+    def forward(self, model, sample, reduce=False):
         """Compute the loss for the given sample.
 
         Returns a tuple with three elements:

Mutant 2710

--- fairseq/criterions/masked_lm.py
+++ fairseq/criterions/masked_lm.py
@@ -69,7 +69,6 @@
         }
         return loss, sample_size, logging_output
 
-    @staticmethod
     def reduce_metrics(logging_outputs) -> None:
         """Aggregate logging outputs from data parallel training."""
         loss_sum = sum(log.get('loss', 0) for log in logging_outputs)

Mutant 2711

--- fairseq/criterions/masked_lm.py
+++ fairseq/criterions/masked_lm.py
@@ -78,7 +78,6 @@
         metrics.log_scalar('loss', loss_sum / sample_size / math.log(2), sample_size, round=3)
         metrics.log_derived('ppl', lambda meters: utils.get_perplexity(meters['loss'].avg))
 
-    @staticmethod
     def logging_outputs_can_be_summed() -> bool:
         """
         Whether the logging outputs returned by `forward` can be summed