KaneAI - World’s First E2E Software Testing Agent.Book a Demo Join TestMu Conference

How to use reporter_class method in Slash

Best Python code snippet using slash

run_experiment.py

Source:run_experiment.py

...245  model_class = choose_model_class(args)246  regimen_class = regimen.ProbeRegimen247  task = task_class(args)248  expt_dataset = dataset_class(args, task)249  expt_reporter = reporter_class(args, expt_dataset)250  expt_probe = probe_class(args)251  expt_model = model_class(args)252  expt_regimen = regimen_class(args)253  expt_loss = loss_class(args)254  if train_probe:255    print('Training probe...')256    run_train_probe(args, expt_probe, expt_dataset, expt_model, expt_loss, expt_reporter, expt_regimen)257  if report_results:258    print('Reporting results of trained probe...')259    run_report_results(args, expt_probe, expt_dataset, expt_model, expt_loss, expt_reporter, expt_regimen)260def setup_new_experiment_dir(args, yaml_args, reuse_results_path, embeddings_path):261  """Constructs a directory in which results and params will be stored.262  If reuse_results_path is not None, then it is reused; no new263  directory is constrcted....

configuration.py

Source:configuration.py

...217                                       ext=extension)218                file_path = osp.join(self.work_dir, filename)219                file_paths.append(file_path)220            modes = [self.mode for i in range(len(file_paths))]221            reporter = reporter_class(file_paths=file_paths, modes=modes,222                                  **self.reporter_partial_kwargs[idx])223            reporters.append(reporter)224        return reporters225    # TODO: remove, not used226    def _gen_work_mapper(self):227        """ """228        work_mapper = self._work_mapper_class(n_workers=self._default_n_workers)229        return work_mapper230    @property231    def reporters(self):232        """ """233        return deepcopy(self._reporters)234    @property235    def work_mapper(self):...

__init__.py

Source:__init__.py

1from argparse import ArgumentParser2import os3import random4from datetime import datetime5import shutil6import yaml7from tqdm import tqdm8import torch9import random10import numpy as np11from . import data, model, probe, regimen, reporter, task, loss12def choose_task_classes(args):13  """Chooses which task class to use based on config.14  Args:15    args: the global config dictionary built by yaml.16  Returns:17    A class to be instantiated as a task specification.18  """19  if args['probe']['task_name'] == 'parse-distance':20    task_class = task.ParseDistanceTask21    reporter_class = reporter.WordPairReporter22    if args['probe_training']['loss'] == 'L1':23      loss_class = loss.L1DistanceLoss24    else:25      raise ValueError("Unknown loss type for given probe type: {}".format(26        args['probe_training']['loss']))27  elif args['probe']['task_name'] == 'parse-depth':28    task_class = task.ParseDepthTask29    reporter_class = reporter.WordReporter30    if args['probe_training']['loss'] == 'L1':31      loss_class = loss.L1DepthLoss32    else:33      raise ValueError("Unknown loss type for given probe type: {}".format(34        args['probe_training']['loss']))35  elif args['probe']['task_name'] == 'part-of-speech':36    task_class = task.PartOfSpeechLabelTask37    reporter_class = reporter.WordReporter38    if args['probe_training']['loss'] == 'cross-entropy':39      loss_class = loss.CrossEntropyLoss40    else:41      raise ValueError("Unknown loss type for given probe type: {}".format(42        args['probe_training']['loss']))43  elif args['probe']['task_name'] == 'rand-prefix-label':44    task_class = task.RandomPrefixLabelTask45    reporter_class = reporter.WordReporter46    if args['probe_training']['loss'] == 'cross-entropy':47      loss_class = loss.CrossEntropyLoss48    else:49      raise ValueError("Unknown loss type for given probe type: {}".format(50        args['probe_training']['loss']))51  elif args['probe']['task_name'] == 'rand-word-label':52    task_class = task.RandomWordLabelTask53    reporter_class = reporter.WordReporter54    if args['probe_training']['loss'] == 'cross-entropy':55      loss_class = loss.CrossEntropyLoss56    else:57      raise ValueError("Unknown loss type for given probe type: {}".format(58        args['probe_training']['loss']))59  elif args['probe']['task_name'] == 'corrupted-part-of-speech':60    task_class = task.CorruptedPartOfSpeechLabelTask61    reporter_class = reporter.WordReporter62    if args['probe_training']['loss'] == 'cross-entropy':63      loss_class = loss.CrossEntropyLoss64    else:65      raise ValueError("Unknown loss type for given probe type: {}".format(66        args['probe_training']['loss']))67  elif args['probe']['task_name'] == 'corrupted-edge-labels':68    task_class = task.CorruptedEdgePositionTask69    reporter_class = reporter.WordPairLabelReporter70    if args['probe_training']['loss'] == 'cross-entropy':71      loss_class = loss.CrossEntropyLoss72    else:73      raise ValueError("Unknown loss type for given probe type: {}".format(74        args['probe_training']['loss']))75      #elif args['probe']['task_name'] == 'corrupted-parse-distance':76      #  task_class = task.CorruptedParseDistanceTask77      #  reporter_class = reporter.WordPairReporter78      #  if args['probe_training']['loss'] == 'L1':79      #    loss_class = loss.L1DistanceLoss80      #  else:81      #    raise ValueError("Unknown loss type for given probe type: {}".format(82      #      args['probe_training']['loss']))83  elif args['probe']['task_name'] == 'rand-parse-depth':84    task_class = task.RandomParseDepthTask85    reporter_class = reporter.WordReporter86    if args['probe_training']['loss'] == 'L1':87      loss_class = loss.L1DepthLoss88    else:89      raise ValueError("Unknown loss type for given probe type: {}".format(90        args['probe_training']['loss']))91  elif args['probe']['task_name'] == 'rand-linear-parse-depth':92    task_class = task.RandomLinearParseDepthTask93    reporter_class = reporter.WordReporter94    if args['probe_training']['loss'] == 'L1':95      loss_class = loss.L1DepthLoss96    else:97      raise ValueError("Unknown loss type for given probe type: {}".format(98        args['probe_training']['loss']))99  elif args['probe']['task_name'] == 'balanced-distance':100    task_class = task.BalancedBinaryTreeDistanceTask101    reporter_class = reporter.WordPairReporter102    if args['probe_training']['loss'] == 'L1':103      loss_class = loss.L1DistanceLoss104    else:105      raise ValueError("Unknown loss type for given probe type: {}".format(106        args['probe_training']['loss']))107  elif args['probe']['task_name'] == 'corrupted-parse-depth':108    task_class = task.CorruptedParseDepthTask109    reporter_class = reporter.WordReporter110    if args['probe_training']['loss'] == 'L1':111      loss_class = loss.L1DepthLoss112    else:113      raise ValueError("Unknown loss type for given probe type: {}".format(114        args['probe_training']['loss']))115  else:116    raise ValueError("Unknown probing task type: {}".format(117      args['probe']['task_name']))118  return task_class, reporter_class, loss_class119def choose_dataset_class(args):120  """Chooses which dataset class to use based on config.121  Args:122    args: the global config dictionary built by yaml.123  Returns:124    A class to be instantiated as a dataset.125  """126  if args['model']['model_type'] in {'ELMo-disk', 'ELMo-random-projection', 'ELMo-decay'}:127    dataset_class = data.ELMoDataset128  elif args['model']['model_type'] == 'BERT-disk':129    dataset_class = data.BERTDataset130  else:131    raise ValueError("Unknown model type for datasets: {}".format(132      args['model']['model_type']))133  return dataset_class134def choose_probe_class(args):135  """Chooses which probe and reporter classes to use based on config.136  Args:137    args: the global config dictionary built by yaml.138  Returns:139    A probe_class to be instantiated.140  """141  if args['probe']['task_signature'] == 'word':142    if args['probe']['psd_parameters']:143      return probe.OneWordPSDProbe144    elif 'probe_spec' not in args['probe'] or args['probe']['probe_spec']['probe_hidden_layers'] == 0:145      return probe.OneWordNonPSDProbe146    else:147      return probe.OneWordNNDepthProbe148  elif args['probe']['task_signature'] == 'word_pair':149    if args['probe']['psd_parameters']:150      return probe.TwoWordPSDProbe151    else:152      return probe.TwoWordNonPSDProbe153  elif args['probe']['task_signature'] == 'word_label':154    if 'probe_spec' not in args['probe'] or args['probe']['probe_spec']['probe_hidden_layers'] == 0:155      return probe.OneWordLinearLabelProbe156    else:157      if 'type' not in args['probe'] or args['probe']['type'] == 'none':158        return probe.OneWordNNLabelProbe159      elif args['probe'] or args['probe']['type'] == 'bayes':160        return probe.OneWordNNLabelProbeBayesCompression161      else:162        raise ValueError("Unknown Probe type: not 'none' and not 'bayesian'")163  elif args['probe']['task_signature'] == 'word_pair_label':164    if args['probe']['probe_spec']['probe_type'] == 'linear':165      return probe.TwoWordLinearLabelProbe166    if args['probe']['probe_spec']['probe_type'] == 'bilinear':167      return probe.TwoWordBilinearLabelProbe168    if args['probe']['probe_spec']['probe_type'] == 'MLP':169      return probe.TwoWordNNLabelProbe170    else:171      raise ValueError("Unknown TwoWordLabel type")172  else:173    raise ValueError("Unknown probe type (probe function signature): {}".format(174      args['probe']['task_signature']))175def choose_model_class(args):176  """Chooses which reporesentation learner class to use based on config.177  Args:178    args: the global config dictionary built by yaml.179  Returns:180    A class to be instantiated as a model to supply word representations.181  """182  if args['model']['model_type'] == 'ELMo-disk':183    return model.DiskModel184  elif args['model']['model_type'] == 'BERT-disk':185    return model.DiskModel186  elif args['model']['model_type'] == 'ELMo-random-projection':187    return model.ProjectionModel188  elif args['model']['model_type'] == 'ELMo-decay':189    return model.DecayModel190  elif args['model']['model_type'] == 'pytorch_model':191    raise ValueError("Using pytorch models for embeddings not yet supported...")192  else:193    raise ValueError("Unknown model type: {}".format(194      args['model']['model_type']))195def run_train_probe(args, probe, dataset, model, loss, reporter, regimen):196  """Trains a structural probe according to args.197  Args:198    args: the global config dictionary built by yaml.199          Describes experiment settings.200    probe: An instance of probe.Probe or subclass.201          Maps hidden states to linguistic quantities.202    dataset: An instance of data.SimpleDataset or subclass.203          Provides access to DataLoaders of corpora. 204    model: An instance of model.Model205          Provides word representations.206    reporter: An instance of reporter.Reporter207          Implements evaluation and visualization scripts.208  Returns:209    None; causes probe parameters to be written to disk.210  """211  regimen.train_until_convergence(probe, model, loss,212      dataset.get_train_dataloader(), dataset.get_dev_dataloader())213def run_report_results(args, probe, dataset, model, loss, reporter, regimen):214  """215  Reports results from a structural probe according to args.216  By default, does so only for dev set.217  Requires a simple code change to run on the test set.218  """219  probe_params_path = os.path.join(args['reporting']['root'],args['probe']['params_path'])220  probe.load_state_dict(torch.load(probe_params_path))221  probe.eval()222  dev_dataloader = dataset.get_dev_dataloader()223  dev_predictions = regimen.predict(probe, model, dev_dataloader)...

reporter.py

Source:reporter.py

...57        """58        [reporter_name, run_type] = extract_embedded_run_type(reporter_name, run_type)59        reporter_config = self.get_reporter_config(reporter_name, scan_type, run_type)60        reporter_class: ReporterMeta = self.reporters[reporter_name]61        reporter_instance = reporter_class(reporter_config)62        return reporter_instance63    def _add_reporters(self, reporters: dict):64        """adds new tools to tools registry"""65        for reporter_name in reporters:66            reporter = reporters[reporter_name]67            if issubclass(reporter, ReporterMeta):68                if not hasattr(self.reporters, reporter_name):69                    log_debug(f"-- installing reporter '{reporter_name}'")70                    self.reporters[reporter_name] = reporter71                else:72                    log_debug(f"-- skipping '{reporter_name}' already defined")73                    continue74            # TODO: else check public functions75            else:...

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.