def reduce_metrics(logging_outputs) -> None: """Aggregate logging outputs from data parallel training.""" loss_sum = utils.item(sum(log.get('loss', 0) for log in logging_outputs)) ntokens = utils.item(sum(log.get('ntokens', 0) for log in logging_outputs)) sample_size = utils.item(sum(log.get('sample_size', 0) for log in logging_outputs)) metrics.log_scalar('loss', loss_sum / sample_size / math.log(2), sample_size, round=3) if sample_size != ntokens: metrics.log_scalar('nll_loss', loss_sum / ntokens / math.log(2), ntokens, round=3) metrics.log_derived('ppl', lambda meters: utils.get_perplexity(meters['nll_loss'].avg)) else: metrics.log_derived('ppl', lambda meters: utils.get_perplexity(meters['loss'].avg))
def reduce_metrics(logging_outputs) -> None: """Aggregate logging outputs from data parallel training.""" sample_size = utils.item( sum(log.get("sample_size", 0) for log in logging_outputs)) loss = utils.item(sum(log.get("loss", 0) for log in logging_outputs)) nll_loss = utils.item( sum(log.get("nll_loss", 0) for log in logging_outputs)) metrics.log_scalar('loss', loss / sample_size / math.log(2), sample_size, round=3) metrics.log_scalar('nll_loss', nll_loss / sample_size / math.log(2), sample_size, round=3) metrics.log_derived( 'ppl', lambda meters: utils.get_perplexity(meters['loss'].avg)) for key in logging_outputs[0]: if key[-5:] == "-loss": val = sum(log.get(key, 0) for log in logging_outputs) metrics.log_scalar( key[:-5], val / sample_size / math.log(2) if sample_size > 0 else 0.0, sample_size, round=3, )
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) sample_size = sum(log.get('sample_size', 0) for log in logging_outputs) 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))