提交

kongdeqiang

5 天以前 28ccfbfc51068a663a80764e14074df5edf2b5ba

 funasr/bin/train_ds.py

@@ -27,7 +27,7 @@
from funasr.train_utils.trainer_ds import Trainer
from funasr.schedulers import scheduler_classes
from funasr.train_utils.initialize import initialize
from funasr.download.download_from_hub import download_model
from funasr.download.download_model_from_hub import download_model
from funasr.models.lora.utils import mark_only_lora_as_trainable
from funasr.train_utils.set_all_random_seed import set_all_random_seed
from funasr.train_utils.load_pretrained_model import load_pretrained_model
@@ -81,8 +81,13 @@
        deepspeed.init_distributed(dist_backend=kwargs.get("backend", "nccl"))
    elif use_ddp or use_fsdp:
        logging.info(f"use_ddp: {use_ddp}, use_fsdp: {use_fsdp}")
        dist.init_process_group(backend=kwargs.get("backend", "nccl"), init_method="env://")
        dist.init_process_group(
            backend=kwargs.get("backend", "nccl"),
            init_method="env://",
        )
        torch.cuda.set_device(local_rank)

    # rank = dist.get_rank()

    logging.info("Build model, frontend, tokenizer")
    device = kwargs.get("device", "cuda")
@@ -124,38 +129,17 @@
        use_ddp=use_ddp,
        use_fsdp=use_fsdp,
        device=kwargs["device"],
        excludes=kwargs.get("excludes", None),
        output_dir=kwargs.get("output_dir", "./exp"),
        **kwargs.get("train_conf"),
    )

    model = trainer.warp_model(model)
    model = trainer.warp_model(model, **kwargs)

    kwargs["device"] = next(model.parameters()).device
    trainer.device = kwargs["device"]
    kwargs["device"] = int(os.environ.get("LOCAL_RANK", 0))
    trainer.device = int(os.environ.get("LOCAL_RANK", 0))

    # optim
    logging.info("Build optim")
    optim = kwargs.get("optim", "adam")
    assert optim in optim_classes
    optim_class = optim_classes.get(optim)
    optim = optim_class(model.parameters(), **kwargs.get("optim_conf"))

    # scheduler
    logging.info("Build scheduler")
    scheduler = kwargs.get("scheduler", "warmuplr")
    assert scheduler in scheduler_classes
    scheduler_class = scheduler_classes.get(scheduler)
    scheduler = scheduler_class(optim, **kwargs.get("scheduler_conf"))

    if use_deepspeed:
        args = OmegaConf.create({"deepspeed_config": kwargs.get("deepspeed_config", "")})
        model, optimizer, _, scheduler = deepspeed.initialize(
            args=args,
            model=model,
            optimizer=optim,
            lr_scheduler=scheduler,
            model_parameters=model.parameters(),
        )
    model, optim, scheduler = trainer.warp_optim_scheduler(model, **kwargs)

    # dataset
    logging.info("Build dataloader")
@@ -165,7 +149,7 @@
    dataloader = dataloader_class(**kwargs)
    # dataloader_tr, dataloader_val = dataloader_class(**kwargs)

    scaler = GradScaler(enabled=trainer.use_fp16) if trainer.use_fp16 else None
    scaler = GradScaler(enabled=True) if trainer.use_fp16 else None
    scaler = ShardedGradScaler(enabled=trainer.use_fp16) if trainer.use_fsdp else scaler

    trainer.resume_checkpoint(
@@ -175,20 +159,17 @@
        scaler=scaler,
    )

    tensorboard_dir = os.path.join(kwargs.get("output_dir"), "tensorboard")
    os.makedirs(tensorboard_dir, exist_ok=True)
    try:
        from tensorboardX import SummaryWriter

        writer = SummaryWriter(tensorboard_dir)  # if trainer.rank == 0 else None
    except:
        writer = None
    early_stopping_patience = kwargs.get("train_conf", {}).get("early_stopping_patience", 0)
    best_val_loss = float("inf")
    epochs_no_improve = 0

    dataloader_tr, dataloader_val = None, None
    for epoch in range(trainer.start_epoch, trainer.max_epoch):
        time1 = time.perf_counter()

        for data_split_i in range(trainer.start_data_split_i, dataloader.data_split_num):
            time_slice_i = time.perf_counter()

            dataloader_tr, dataloader_val = dataloader.build_iter(
                epoch, data_split_i=data_split_i, start_step=trainer.start_step
            )
@@ -201,20 +182,40 @@
                dataloader_train=dataloader_tr,
                dataloader_val=dataloader_val,
                epoch=epoch,
                writer=writer,
                data_split_i=data_split_i,
                data_split_num=dataloader.data_split_num,
                start_step=trainer.start_step,
            )
            trainer.start_step = 0

            torch.cuda.empty_cache()
            device = next(model.parameters()).device
            if device.type == "cuda":
                with torch.cuda.device(device):
                    torch.cuda.empty_cache()

            time_escaped = (time.perf_counter() - time_slice_i) / 3600.0
            logging.info(
                f"\n\nrank: {local_rank}, "
                f"time_escaped_epoch: {time_escaped:.3f} hours, "
                f"estimated to finish {dataloader.data_split_num} data_slices, remaining: {dataloader.data_split_num-data_split_i} slices, {(dataloader.data_split_num-data_split_i)*time_escaped:.3f} hours, "
                f"epoch: {trainer.max_epoch - epoch} epochs, {((trainer.max_epoch - epoch - 1)*dataloader.data_split_num + dataloader.data_split_num-data_split_i)*time_escaped:.3f} hours\n"
            )

        trainer.start_data_split_i = 0
        trainer.validate_epoch(
            model=model, dataloader_val=dataloader_val, epoch=epoch + 1, writer=writer
        )
        scheduler.step()
        trainer.validate_epoch(model=model, dataloader_val=dataloader_val, epoch=epoch + 1)
        current_val = trainer.val_loss_avg

        if current_val < best_val_loss:
            logging.info(f"current_val: {current_val}, best_val_loss: {best_val_loss}")
            best_val_loss = current_val
            epochs_no_improve = 0
        else:
            epochs_no_improve += 1
            logging.info(f"No val_loss improvement for {epochs_no_improve}/{early_stopping_patience} epochs")
        if early_stopping_patience > 0 and epochs_no_improve >= early_stopping_patience:
            logging.info(f"Early stopping triggered at epoch {epoch+1}")
            break

        trainer.step_in_epoch = 0
        trainer.save_checkpoint(
            epoch + 1, model=model, optim=optim, scheduler=scheduler, scaler=scaler
@@ -223,7 +224,7 @@
        time2 = time.perf_counter()
        time_escaped = (time2 - time1) / 3600.0
        logging.info(
            f"rank: {local_rank}, "
            f"\n\nrank: {local_rank}, "
            f"time_escaped_epoch: {time_escaped:.3f} hours, "
            f"estimated to finish {trainer.max_epoch} "
            f"epoch: {(trainer.max_epoch - epoch) * time_escaped:.3f} hours\n"
@@ -232,7 +233,9 @@
        trainer.train_loss_avg = 0.0

    if trainer.rank == 0:
        average_checkpoints(trainer.output_dir, trainer.avg_nbest_model)
        average_checkpoints(
            trainer.output_dir, trainer.avg_nbest_model, use_deepspeed=trainer.use_deepspeed
        )

    trainer.close()