70b15bde1d8a1b29d24f23bac1a28a63be0fb4d2,torchsample/modules/super_module.py,SuperModule,evaluate_loader,#SuperModule#Any#,223

Before Change

    def evaluate_loader(self, loader):
        nb_batches = int(math.ceil(len(loader.dataset.inputs)/loader.batch_size))
        losses = torch.FloatTensor(nb_batches)
        for batch_idx, (x_batch, y_batch) in enumerate(loader):
            x_batch = Variable(x_batch)
            y_batch = Variable(y_batch)

            y_pred = self(x_batch)
            loss = self._loss(y_pred, y_batch)
            losses[batch_idx] = loss.data[0]

        return torch.mean(losses)

    def save(self, file):
        
        Save a model to disk

After Change

            y_pred = self(x_batch)
            loss = self._loss(y_pred, y_batch)
            total_loss += loss.data[0]*len(x_batch)
            total_samples += len(x_batch)
        self.train()
        return total_loss / total_samples

    def evaluate_on_batch(self, x, y):
        self.eval()
        x = Variable(x)

Instances