train.py

import config
import utils
import torch.nn as nn
import torch.optim as optim
import numpy as np

from tqdm import tqdm
from torch.utils.data import DataLoader
from dataset import CollateDataset


def train_epoch(loader, model, optimizer, loss_fn, epoch):
    model.train()

    losses = []

    loader = tqdm(loader)

    for img, captions in loader:
        img = img.to(config.DEVICE)
        captions = captions.to(config.DEVICE)

        output = model(img, captions)

        loss = loss_fn(
            output.reshape(-1, output.shape[2]),
            captions[:, 1:].reshape(-1)
        )

        optimizer.zero_grad()
        loss.backward()
        optimizer.step()

        loader.set_postfix(loss=loss.item())

        losses.append(loss.item())

    if config.SAVE_MODEL:
        utils.save_checkpoint({
            'state_dict': model.state_dict(),
            'optimizer': optimizer.state_dict(),
            'epoch': epoch,
            'loss': np.mean(losses)
        })

    print(f'Epoch[{epoch}]: Loss {np.mean(losses)}')


def main():
    all_dataset = utils.load_dataset()

    train_dataset, _ = utils.train_test_split(dataset=all_dataset)

    train_loader = DataLoader(
        dataset=train_dataset,
        batch_size=config.BATCH_SIZE,
        pin_memory=config.PIN_MEMORY,
        drop_last=False,
        shuffle=True,
        collate_fn=CollateDataset(pad_idx=all_dataset.vocab.stoi['<PAD>']),
    )

    model = utils.get_model_instance(all_dataset.vocab)

    optimizer = optim.Adam(model.parameters(), lr=config.LEARNING_RATE)
    loss_fn = nn.CrossEntropyLoss(ignore_index=all_dataset.vocab.stoi['<PAD>'])

    starting_epoch = 1

    if utils.can_load_checkpoint():
        starting_epoch = utils.load_checkpoint(model, optimizer)

    for epoch in range(starting_epoch, config.EPOCHS):
        train_epoch(
            train_loader,
            model,
            optimizer,
            loss_fn,
            epoch
        )


if __name__ == '__main__':
    main()