How to use test_l2 method in avocado

Best Python code snippet using avocado_python

trainer.py

Source:trainer.py

1"""2@author: Yong Zheng Ong3implements the training procedure4"""5from timeit import default_timer6import os7from matplotlib.pyplot import ylim8import matplotlib.pyplot as plt9import torch10import torch.nn as nn11import numpy as np12from .model import Model13from ...loss.loss import LpLoss14class Trainer():15    def __init__(self, modelConfig, dataset, prefix="default"):16        """17        the trainer object for performing training, testing, etc18        """19        # save dataset20        self.dataset = dataset21        self.prefix = prefix22        self.save_folder = "./results/iae_net_{}.pt".format(self.prefix)23        # add additional dataset related details to modelConfig24        modelConfig["num_d"] = self.dataset.num_d25        modelConfig["input_channel"] = self.dataset.input_channel26        modelConfig["output_channel"] = self.dataset.output_channel27        modelConfig["size"] = self.dataset.s28        print(modelConfig)29        # instantiate model30        self.model = Model(**modelConfig).cuda()31        print("model will be saved to.. {}".format(self.save_folder))32    def train(self, epochs, batch_size, learning_rate, gamma):33        """34        perform training of the model for n epochs35        """36        available_subs = self.dataset.available_subs37        best_results = {38            'epoch': 0,39            'train': 0,40            'test': 10000000041        }42        if self.dataset.load_type in ["default", "multi"]:43            if self.dataset.y_train is None: # already packaged into dataset class44                train_loader = torch.utils.data.DataLoader(self.dataset.x_train, batch_size=batch_size, shuffle=True)45                test_loader = torch.utils.data.DataLoader(self.dataset.x_test, batch_size=batch_size, shuffle=False)46            else:47                train_loader = torch.utils.data.DataLoader(torch.utils.data.TensorDataset(self.dataset.x_train, self.dataset.y_train), batch_size=batch_size, shuffle=True)48                test_loader = torch.utils.data.DataLoader(torch.utils.data.TensorDataset(self.dataset.x_test, self.dataset.y_test), batch_size=batch_size, shuffle=False)49        elif self.dataset.load_type in ["multi3"]:50            train_loader = torch.utils.data.DataLoader(torch.utils.data.TensorDataset(*self.dataset.x_train, *self.dataset.y_train), batch_size=batch_size, shuffle=True)51            test_loader = torch.utils.data.DataLoader(torch.utils.data.TensorDataset(*self.dataset.x_test, *self.dataset.y_test), batch_size=batch_size, shuffle=False)52        optimizer = torch.optim.Adam(self.model.parameters(), lr=learning_rate, weight_decay=1e-4)53        scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(optimizer, factor=gamma, patience=20)54        myloss = LpLoss(size_average=False)55        for ep in range(epochs):56            self.model.train()57            t1 = default_timer()58            train_l2 = 0.059            if self.dataset.load_type in ["default", "multi"]:60                for x, y in train_loader:61                    optimizer.zero_grad()62                    if self.dataset.num_d == 1:63                        if self.dataset.load_type == "default":64                            x, y = x.cuda(), y.cuda()65                            out = self.model(x)66                            l2 = myloss(out.reshape(batch_size, -1), y.reshape(batch_size, -1))67                            loss = l268                            loss.backward()69                        elif self.dataset.load_type == "multi":70                            for i in range(len(available_subs)):71                                # subsample the training data72                                x_temp = x[:,::available_subs[i]].cuda()73                                y_temp = y[:,::available_subs[i]].cuda()74                                out = self.model(x_temp)75                                loss = myloss(out.reshape(batch_size, -1), y_temp.reshape(batch_size, -1))76                                loss.backward()77                    if self.dataset.num_d == 2:78                        if self.dataset.load_type == "multi":79                            for i in range(len(available_subs)):80                                # subsample the training data81                                x_temp, y_temp = self.dataset.build_grid_data(x, y, i)82                                x_temp = x_temp.cuda()83                                y_temp = y_temp.cuda()84                                out = self.model(x_temp).squeeze()85                                if self.dataset.num_d == 2 and self.dataset.y_normalizer is not None:86                                    out = self.dataset.y_normalizer[i].decode(out)87                                    y_temp = self.dataset.y_normalizer[i].decode(y_temp)88                                loss = myloss(out.reshape(batch_size, -1), y_temp.reshape(batch_size, -1))89                                loss.backward()90                    optimizer.step()91                    train_l2 += loss.item()92            elif self.dataset.load_type in ["multi3"]:93                for data in train_loader:94                    optimizer.zero_grad()95                    loss = 0.096                    for i in range(len(available_subs)):97                        x = data[i].cuda()98                        y = data[len(available_subs)+i].cuda()99                        out = self.model(x)100                        if self.dataset.num_d == 2 and self.dataset.y_normalizer is not None:101                            out = self.dataset.y_normalizer[i].decode(out)102                            y = self.dataset.y_normalizer[i].decode(y)103                        loss += myloss(out.reshape(batch_size, -1), y.reshape(batch_size, -1))104                    loss.backward()105                    optimizer.step()106                    train_l2 += loss.item()107            train_l2 /= self.dataset.ntrain108            # test the model109            self.model.eval()110            test_l2 = np.array([0.0] * len(available_subs))111            with torch.no_grad():112                if self.dataset.load_type in ["default", "multi"]:113                    for x, y in test_loader:114                        x, y = x.cuda(), y.cuda()115                        if self.dataset.num_d == 1:116                            if self.dataset.load_type in ["default", "multi"]:117                                for i in range(len(available_subs)):118                                    # subsample the training data119                                    x_temp = x[:,::available_subs[i]]120                                    y_temp = y[:,::available_subs[i]]121                                    out = self.model(x_temp)122                                    test_l2[i] += myloss(out.reshape(batch_size, -1), y_temp.reshape(batch_size, -1)).item()123                        if self.dataset.num_d == 2:124                            if self.dataset.load_type in ["default", "multi"]:125                                for i in range(len(available_subs)):126                                    # subsample the training data127                                    x_temp, y_temp = self.dataset.build_grid_data(x, y, i)128                                    out = self.model(x_temp).squeeze()129                                    if self.dataset.num_d == 2 and self.dataset.y_normalizer is not None:130                                        out = self.dataset.y_normalizer[i].decode(out)131                                        y_temp = self.dataset.y_normalizer[i].decode(y_temp)132                                    test_l2[i] += myloss(out.reshape(batch_size, -1), y_temp.reshape(batch_size, -1)).item()133                elif self.dataset.load_type in ["multi3"]:134                    for data in test_loader:135                        for i in range(len(available_subs)):136                            x = data[i].cuda()137                            y = data[len(available_subs)+i].cuda()138                            out = self.model(x).squeeze()139                            if self.dataset.num_d == 2 and self.dataset.y_normalizer is not None:140                                out = self.dataset.y_normalizer[i].decode(out)141                                y = self.dataset.y_normalizer[i].decode(y)142                            test_l2[i] += myloss(out.reshape(batch_size, -1), y.reshape(batch_size, -1)).item()143            test_l2 /= self.dataset.ntest144            scheduler.step(test_l2[self.dataset.ts])145            t2 = default_timer()146            # if results improved, update best results147            if self.dataset.num_d == 1 and test_l2[self.dataset.ts] <= best_results['test']:148                print("saving improved model...")149                best_results['epoch'] = ep150                best_results['train'] = train_l2151                best_results['test'] = test_l2[self.dataset.ts]152                torch.save(self.model, self.save_folder)153            if self.dataset.num_d == 2 and test_l2[self.dataset.ts] <= best_results['test']:154                print("saving improved model...")155                best_results['epoch'] = ep156                best_results['train'] = train_l2157                best_results['test'] = test_l2[self.dataset.ts]158                torch.save(self.model, self.save_folder)159            print(ep, f'{t2-t1:.4e}', f'{train_l2:.4e}', test_l2, scheduler.optimizer.param_groups[0]['lr'])...

test_multipler.py

Source:test_multipler.py

1import pytest2from sandpit_package.mathers.multiplier import Multiplier3expected_op = "*"4test_l1 = 55test_r1 = 76expected_result1 = test_l1 * test_r17expected_msg1 = "%s %s %s = %s" % (test_l1, expected_op, test_r1, expected_result1)8test_l2 = "abc"9test_r2 = 510expected_result2 = test_l2 * test_r211expected_msg2 = "%s %s %s = %s" % (test_l2, expected_op, test_r2, expected_result2)12def test_init():13    with pytest.raises(Exception):14        x1 = Multiplier()15    with pytest.raises(Exception):16        x2 = Multiplier(test_l1)17    x3 = Multiplier(test_l1, test_r1)18    assert (x3.l, x3.r) == (test_l1, test_r1)19    x4 = Multiplier(test_l2, test_r2)20    assert (x4.l, x4.r) == (test_l2, test_r2)21def test_op_str():22    x = Multiplier(test_l1, test_r1)23    assert x._op_str() == expected_op24def test_result():25    x1 = Multiplier(test_l1, test_r1)26    assert x1.result() == expected_result127    x1 = Multiplier(test_r1, test_l1)28    assert x1.result() == expected_result129    x2 = Multiplier(test_l2, test_r2)30    assert x2.result() == expected_result231    x2 = Multiplier(test_r2, test_l2)32    assert x2.result() == expected_result233def test_msg():34    x1 = Multiplier(test_l1, test_r1)35    assert x1._msg() == expected_msg136    x2 = Multiplier(test_l2, test_r2)37    assert x2._msg() == expected_msg238def test_print(capsys):39    x1 = Multiplier(test_l1, test_r1)40    x1.print()...

test_adder.py

Source:test_adder.py

1import pytest2from sandpit_package.mathers.adder import Adder3expected_op = "+"4test_l1 = 55test_r1 = 76expected_result1 = test_l1 + test_r17expected_msg1 = "%s %s %s = %s" % (test_l1, expected_op, test_r1, expected_result1)8test_l2 = "abc"9test_r2 = "def"10expected_result2 = test_l2 + test_r211expected_msg2 = "%s %s %s = %s" % (test_l2, expected_op, test_r2, expected_result2)12def test_init():13    with pytest.raises(Exception):14        x1 = Adder()15    with pytest.raises(Exception):16        x2 = Adder(test_l1)17    x3 = Adder(test_l1, test_r1)18    assert (x3.l, x3.r) == (test_l1, test_r1)19    x4 = Adder(test_l2, test_r2)20    assert (x4.l, x4.r) == (test_l2, test_r2)21def test_op_str():22    x = Adder(test_l1, test_r1)23    assert x._op_str() == expected_op24def test_result():25    x1 = Adder(test_l1, test_r1)26    assert x1.result() == expected_result127    x1 = Adder(test_r1, test_l1)28    assert x1.result() == expected_result129    x2 = Adder(test_l2, test_r2)30    assert x2.result() == expected_result231    x2 = Adder(test_r2, test_l2)32    assert x2.result() != expected_result233def test_msg():34    x1 = Adder(test_l1, test_r1)35    assert x1._msg() == expected_msg136    x2 = Adder(test_l2, test_r2)37    assert x2._msg() == expected_msg238def test_print(capsys):39    x1 = Adder(test_l1, test_r1)40    x1.print()...

Automation Testing Tutorials

Learn to execute automation testing from scratch with LambdaTest Learning Hub. Right from setting up the prerequisites to run your first automation test, to following best practices and diving deeper into advanced test scenarios. LambdaTest Learning Hubs compile a list of step-by-step guides to help you be proficient with different test automation frameworks i.e. Selenium, Cypress, TestNG etc.