How to use main_loop method in ATX

Best Python code snippet using ATX

test_training.py

Source:test_training.py

1from tempfile import NamedTemporaryFile2import numpy3from numpy.testing import assert_allclose4import theano5from fuel.datasets import IterableDataset6from theano import tensor7from blocks.algorithms import GradientDescent, Scale8from blocks.config import config9from blocks.extensions import FinishAfter, TrainingExtension10from blocks.extensions.saveload import Checkpoint11from blocks.extensions.training import SharedVariableModifier, TrackTheBest12from blocks.extensions.predicates import OnLogRecord13from blocks.main_loop import MainLoop14from blocks.serialization import load15from blocks.utils import shared_floatx16from blocks.utils.testing import MockMainLoop, skip_if_configuration_set17def test_shared_variable_modifier():18    weights = numpy.array([-1, 1], dtype=theano.config.floatX)19    features = [numpy.array(f, dtype=theano.config.floatX)20                for f in [[1, 2], [3, 4], [5, 6]]]21    targets = [(weights * f).sum() for f in features]22    n_batches = 323    dataset = IterableDataset(dict(features=features, targets=targets))24    x = tensor.vector('features')25    y = tensor.scalar('targets')26    W = shared_floatx([0, 0], name='W')27    cost = ((x * W).sum() - y) ** 228    cost.name = 'cost'29    step_rule = Scale(0.001)30    sgd = GradientDescent(cost=cost, parameters=[W],31                          step_rule=step_rule)32    main_loop = MainLoop(33        model=None, data_stream=dataset.get_example_stream(),34        algorithm=sgd,35        extensions=[36            FinishAfter(after_n_epochs=1),37            SharedVariableModifier(38                step_rule.learning_rate,39                lambda n: numpy.cast[theano.config.floatX](10. / n)40            )])41    main_loop.run()42    assert_allclose(step_rule.learning_rate.get_value(),43                    numpy.cast[theano.config.floatX](10. / n_batches))44def test_shared_variable_modifier_two_parameters():45    weights = numpy.array([-1, 1], dtype=theano.config.floatX)46    features = [numpy.array(f, dtype=theano.config.floatX)47                for f in [[1, 2], [3, 4], [5, 6]]]48    targets = [(weights * f).sum() for f in features]49    n_batches = 350    dataset = IterableDataset(dict(features=features, targets=targets))51    x = tensor.vector('features')52    y = tensor.scalar('targets')53    W = shared_floatx([0, 0], name='W')54    cost = ((x * W).sum() - y) ** 255    cost.name = 'cost'56    step_rule = Scale(0.001)57    sgd = GradientDescent(cost=cost, parameters=[W],58                          step_rule=step_rule)59    modifier = SharedVariableModifier(60        step_rule.learning_rate,61        lambda _, val: numpy.cast[theano.config.floatX](val * 0.2))62    main_loop = MainLoop(63        model=None, data_stream=dataset.get_example_stream(),64        algorithm=sgd,65        extensions=[FinishAfter(after_n_epochs=1), modifier])66    main_loop.run()67    new_value = step_rule.learning_rate.get_value()68    assert_allclose(new_value,69                    0.001 * 0.2 ** n_batches,70                    atol=1e-5)71def test_track_the_best():72    main_loop = MockMainLoop()73    extension = TrackTheBest("cost")74    extension.main_loop = main_loop75    main_loop.status['epochs_done'] += 176    main_loop.status['iterations_done'] += 1077    main_loop.log.current_row['cost'] = 578    extension.dispatch('after_epoch')79    assert main_loop.status['best_cost'] == 580    assert main_loop.log.current_row['cost_best_so_far']81    main_loop.status['epochs_done'] += 182    main_loop.status['iterations_done'] += 1083    main_loop.log.current_row['cost'] = 684    extension.dispatch('after_epoch')85    assert main_loop.status['best_cost'] == 586    assert main_loop.log.current_row.get('cost_best_so_far', None) is None87    main_loop.status['epochs_done'] += 188    main_loop.status['iterations_done'] += 1089    main_loop.log.current_row['cost'] = 590    extension.dispatch('after_epoch')91    assert main_loop.status['best_cost'] == 592    assert main_loop.log.current_row.get('cost_best_so_far', None) is None93    main_loop.status['epochs_done'] += 194    main_loop.status['iterations_done'] += 1095    main_loop.log.current_row['cost'] = 496    extension.dispatch('after_epoch')97    assert main_loop.status['best_cost'] == 498    assert main_loop.log.current_row['cost_best_so_far']99class WriteCostExtension(TrainingExtension):100    def after_batch(self, batch):101        self.main_loop.log.current_row['cost'] = abs(102            self.main_loop.log.status['iterations_done'] - 5) + 3103def test_save_the_best():104    skip_if_configuration_set('log_backend', 'sqlite',105                              "Known to be flaky with SQLite log backend.")106    with NamedTemporaryFile(dir=config.temp_dir) as dst,\107            NamedTemporaryFile(dir=config.temp_dir) as dst_best:108        track_cost = TrackTheBest("cost", after_epoch=False, after_batch=True)109        main_loop = MockMainLoop(110            extensions=[FinishAfter(after_n_epochs=1),111                        WriteCostExtension(),112                        track_cost,113                        Checkpoint(dst.name, after_batch=True,114                                   save_separately=['log'])115                        .add_condition(116                            ["after_batch"],117                            OnLogRecord(track_cost.notification_name),118                            (dst_best.name,))])119        main_loop.run()120        assert main_loop.log[4]['saved_to'] == (dst.name, dst_best.name)121        assert main_loop.log[5]['saved_to'] == (dst.name, dst_best.name)122        assert main_loop.log[6]['saved_to'] == (dst.name,)123        with open(dst_best.name, 'rb') as src:...

connection.py

Source:connection.py

1# SPDX-License-Identifier: BSD-3-Clause2# Handle NM.RemoteConnection3from __future__ import absolute_import, division, print_function4__metaclass__ = type5import logging6# Relative import is not support by ansible 2.8 yet7# pylint: disable=import-error, no-name-in-module8from ansible_collections.fedora.linux_system_roles.plugins.module_utils.network_lsr.nm import client  # noqa:E5019from ansible_collections.fedora.linux_system_roles.plugins.module_utils.network_lsr.nm import error  # noqa:E50110# pylint: enable=import-error, no-name-in-module11def delete_remote_connection(nm_profile, timeout, check_mode):12    if not nm_profile:13        logging.info("NULL NM.RemoteConnection, no need to delete")14        return False15    if not check_mode:16        main_loop = client.get_mainloop(timeout)17        user_data = main_loop18        nm_profile.delete_async(19            main_loop.cancellable,20            _nm_profile_delete_call_back,21            user_data,22        )23        logging.debug(24            "Deleting profile %s/%s with timeout %s",25            nm_profile.get_id(),26            nm_profile.get_uuid(),27            timeout,28        )29        main_loop.run()30    return True31def _nm_profile_delete_call_back(nm_profile, result, user_data):32    main_loop = user_data33    if main_loop.is_cancelled:34        return35    try:36        success = nm_profile.delete_finish(result)37    except Exception as e:38        main_loop.fail(39            error.LsrNetworkNmError(40                "Connection deletion aborted on {id}/{uuid}: error={error}".format(41                    id=nm_profile.get_id(), uuid=nm_profile.get_uuid(), error=e42                )43            )44        )45    if success:46        main_loop.quit()47    else:48        main_loop.fail(49            error.LsrNetworkNmError(50                "Connection deletion aborted on {id}/{uuid}: error=unknown".format(51                    id=nm_profile.get_id(), uuid=nm_profile.get_uuid()52                )53            )54        )55def volatilize_remote_connection(nm_profile, timeout, check_mode):56    if not nm_profile:57        logging.info("NULL NM.RemoteConnection, no need to volatilize")58        return False59    if not check_mode:60        main_loop = client.get_mainloop(timeout)61        user_data = main_loop62        nm_profile.update2(63            None,  # settings64            client.NM.SettingsUpdate2Flags.IN_MEMORY_ONLY65            | client.NM.SettingsUpdate2Flags.VOLATILE,66            None,  # args67            main_loop.cancellable,68            _nm_profile_volatile_update2_call_back,69            user_data,70        )71        logging.debug(72            "Volatilizing profile %s/%s with timeout %s",73            nm_profile.get_id(),74            nm_profile.get_uuid(),75            timeout,76        )77        main_loop.run()78    return True79def _nm_profile_volatile_update2_call_back(nm_profile, result, user_data):80    main_loop = user_data81    if main_loop.is_cancelled:82        return83    try:84        success = nm_profile.update2_finish(result)85    except Exception as e:86        main_loop.fail(87            error.LsrNetworkNmError(88                "Connection volatilize aborted on {id}/{uuid}: error={error}".format(89                    id=nm_profile.get_id(), uuid=nm_profile.get_uuid(), error=e90                )91            )92        )93    if success:94        main_loop.quit()95    else:96        main_loop.fail(97            error.LsrNetworkNmError(98                "Connection volatilize aborted on {id}/{uuid}: error=unknown".format(99                    id=nm_profile.get_id(), uuid=nm_profile.get_uuid()100                )101            )...

test_main_loop.py

Source:test_main_loop.py

...16    """Writes data saved by MockAlgorithm to the log."""17    def after_batch(self, _):18        self.main_loop.log.current_row['batch'] = \19            self.main_loop.algorithm.batch20def test_main_loop():21    old_config_profile_value = config.profile22    config.profile = True23    main_loop = MainLoop(24        MockAlgorithm(), IterableDataset(range(10)).get_example_stream(),25        extensions=[WriteBatchExtension(), FinishAfter(after_n_epochs=2)])26    main_loop.run()27    assert_raises(AttributeError, getattr, main_loop, 'model')28    assert main_loop.log.status['iterations_done'] == 2029    assert main_loop.log.status['_epoch_ends'] == [10, 20]30    assert len(main_loop.log) == 2031    for i in range(20):32        assert main_loop.log[i + 1]['batch'] == {'data': i % 10}33    config.profile = old_config_profile_value34def test_training_resumption():...

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.