def validate()

in bring-your-own-container/fairseq_translation/fairseq/train_driver.py [0:0]

• 41 lines of code
• 8 McCabe index (conditional complexity)

def validate(args, trainer, task, epoch_itr, subsets):
    """Evaluate the model on the validation set(s) and return the losses."""
    valid_losses = []
    for subset in subsets:
        # Initialize data iterator
        itr = task.get_batch_iterator(
            dataset=task.dataset(subset),
            max_tokens=args.max_tokens,
            max_sentences=args.max_sentences_valid,
            max_positions=utils.resolve_max_positions(
                task.max_positions(),
                trainer.get_model().max_positions(),
            ),
            ignore_invalid_inputs=args.skip_invalid_size_inputs_valid_test,
            required_batch_size_multiple=8,
            seed=args.seed,
            num_shards=args.distributed_world_size,
            shard_id=args.distributed_rank,
        ).next_epoch_itr(shuffle=False)
        progress = progress_bar.build_progress_bar(
            args,
            itr,
            epoch_itr.epoch,
            prefix="valid on '{}' subset".format(subset),
            no_progress_bar="simple",
        )

        # reset validation loss meters
        for k in ["valid_loss", "valid_nll_loss"]:
            meter = trainer.get_meter(k)
            if meter is not None:
                meter.reset()
        extra_meters = collections.defaultdict(lambda: AverageMeter())

        for sample in progress:
            log_output = trainer.valid_step(sample)

            for k, v in log_output.items():
                if k in ["loss", "nll_loss", "ntokens", "nsentences", "sample_size"]:
                    continue
                extra_meters[k].update(v)

        # log validation stats
        stats = get_valid_stats(trainer)
        for k, meter in extra_meters.items():
            stats[k] = meter.avg
        progress.print(stats)

        valid_losses.append(stats["valid_loss"])
    return valid_losses