Source code for models.sgd

"""
This module implements the simplest form of incremental training, i.e., finetuning.
"""

from models import register_model
from models.utils.continual_model import ContinualModel



@register_model('sgd')
class Sgd(ContinualModel):
    """
    Finetuning baseline - simple incremental training.
    """

    NAME = 'sgd'
    COMPATIBILITY = ['class-il', 'domain-il', 'task-il', 'general-continual']

    def __init__(self, backbone, loss, args, transform, dataset=None):
        super(Sgd, self).__init__(backbone, loss, args, transform, dataset=dataset)



    def observe(self, inputs, labels, not_aug_inputs, epoch=None):
        """
        SGD trains on the current task using the data provided, with no countermeasures to avoid forgetting.
        """
        self.opt.zero_grad()
        outputs = self.net(inputs)
        loss = self.loss(outputs, labels)
        loss.backward()
        self.opt.step()

        return loss.item()