How to use from_args method in dbt-osmosis

Best Python code snippet using dbt-osmosis_python
mobilenet_edgetpu_v2_model_blocks.py
Source:mobilenet_edgetpu_v2_model_blocks.py
...84      fused_project = False85      conv_type = 'group'86    else:87      raise NotImplementedError(f'Unsupported IBN type {block_op_type.type}.')88    return cls.from_args(89        input_filters=input_filters,90        output_filters=output_filters,91        kernel_size=block_search_config.kernel_size,92        num_repeat=num_repeat,93        expand_ratio=block_search_config.expand_ratio,94        strides=(block_search_config.stride, block_search_config.stride),95        se_ratio=se_ratio,96        id_skip=id_skip,97        fused_expand=fused_expand,98        fused_project=fused_project,99        conv_type=conv_type,100        group_size=block_search_config.group_size)101@dataclasses.dataclass102class BlockGroupConfig(base_config.Config):103  """Config for group of blocks that share the same filter size."""104  blocks: List[BlockSearchConfig] = dataclasses.field(default_factory=list)105  filters: int = 64106def _default_mobilenet_edgetpu_v2_topology():107  return [108      # Block Group 0109      BlockGroupConfig(110          blocks=[111              # BlockSearchConfig: op_type, kernel_size, expand_ratio, stride112              BlockSearchConfig.from_args(113                  BlockType.from_args('ibn_fused'), 3, 1, 1),114          ],115          filters=24),116      # Block Group 1117      BlockGroupConfig(118          blocks=[119              BlockSearchConfig.from_args(120                  BlockType.from_args('ibn_fused'), 3, 8, 2),121              BlockSearchConfig.from_args(122                  BlockType.from_args('ibn_fused_grouped'), 3, 4, 1),123          ],124          filters=48),125      # Block Group 2126      BlockGroupConfig(127          blocks=[128              BlockSearchConfig.from_args(129                  BlockType.from_args('ibn_fused'), 3, 8, 2),130              BlockSearchConfig.from_args(131                  BlockType.from_args('ibn_fused_grouped'), 3, 4, 1),132              BlockSearchConfig.from_args(133                  BlockType.from_args('ibn_fused'), 3, 4, 1),134              BlockSearchConfig.from_args(135                  BlockType.from_args('ibn_fused_grouped'), 3, 4, 1),136          ],137          filters=64),138      # Block Group 3139      BlockGroupConfig(140          blocks=[141              BlockSearchConfig.from_args(142                  BlockType.from_args('ibn_fused'), 3, 8, 2),143              BlockSearchConfig.from_args(144                  BlockType.from_args('ibn_dw'), 3, 4, 1),145              BlockSearchConfig.from_args(146                  BlockType.from_args('ibn_dw'), 3, 4, 1),147              BlockSearchConfig.from_args(148                  BlockType.from_args('ibn_dw'), 3, 4, 1),149          ],150          filters=128),151      # Block Group 4152      BlockGroupConfig(153          blocks=[154              BlockSearchConfig.from_args(155                  BlockType.from_args('ibn_dw'), 3, 8, 1),156              BlockSearchConfig.from_args(157                  BlockType.from_args('ibn_dw'), 3, 4, 1),158              BlockSearchConfig.from_args(159                  BlockType.from_args('ibn_dw'), 3, 4, 1),160              BlockSearchConfig.from_args(161                  BlockType.from_args('ibn_dw'), 3, 4, 1),162          ],163          filters=160),164      # Block Group 5165      BlockGroupConfig(166          blocks=[167              BlockSearchConfig.from_args(168                  BlockType.from_args('ibn_dw'), 3, 8, 2),169              BlockSearchConfig.from_args(170                  BlockType.from_args('ibn_dw'), 3, 4, 1),171              BlockSearchConfig.from_args(172                  BlockType.from_args('ibn_dw'), 3, 4, 1),173              BlockSearchConfig.from_args(174                  BlockType.from_args('ibn_dw'), 3, 4, 1),175          ],176          filters=192),177      # Block Group 6178      BlockGroupConfig(179          blocks=[180              BlockSearchConfig.from_args(181                  BlockType.from_args('ibn_dw'), 3, 8, 1),182          ],183          filters=256),184  ]185@dataclasses.dataclass186class TopologyConfig(base_config.Config):187  """Config for model topology as a collection of BlockGroupConfigs."""188  block_groups: List[BlockGroupConfig] = dataclasses.field(189      default_factory=_default_mobilenet_edgetpu_v2_topology)190@dataclasses.dataclass191class ModelConfig(base_config.Config):192  """Default Config for MobilenetEdgeTPUV2."""193  width_coefficient: float = 1.0194  depth_coefficient: float = 1.0195  resolution: Union[int, Tuple[int, int]] = 224196  dropout_rate: float = 0.1197  stem_base_filters: int = 64198  stem_kernel_size: int = 5199  top_base_filters: int = 1280200  blocks: Tuple[BlockConfig, ...] = (201      # (input_filters, output_filters, kernel_size, num_repeat,202      #  expand_ratio, strides, se_ratio, id_skip, fused_conv, conv_type)203      # pylint: disable=bad-whitespace204      BlockConfig.from_args(205          stem_base_filters, 24, 3, 1, 1, (1, 1), conv_type='full'),206      BlockConfig.from_args(207          24, 48, 3, 1, 8, (2, 2), fused_expand=True, conv_type='full'),208      BlockConfig.from_args(209          48, 48, 3, 1, 4, (1, 1), fused_expand=True, conv_type='group'),210      BlockConfig.from_args(211          48, 64, 3, 1, 8, (2, 2), fused_expand=True, conv_type='full'),212      BlockConfig.from_args(213          64, 64, 3, 1, 4, (1, 1), fused_expand=True, conv_type='group'),214      BlockConfig.from_args(215          64, 64, 3, 1, 4, (1, 1), fused_expand=True, conv_type='full'),216      BlockConfig.from_args(217          64, 64, 3, 1, 4, (1, 1), fused_expand=True, conv_type='group'),218      BlockConfig.from_args(219          64, 128, 3, 1, 8, (2, 2), fused_expand=True, conv_type='full'),220      BlockConfig.from_args(128, 128, 3, 3, 4, (1, 1)),221      BlockConfig.from_args(128, 160, 3, 1, 8, (1, 1)),222      BlockConfig.from_args(160, 160, 3, 3, 4, (1, 1)),223      BlockConfig.from_args(160, 192, 5, 1, 8, (2, 2)),224      BlockConfig.from_args(192, 192, 5, 3, 4, (1, 1)),225      BlockConfig.from_args(192, 256, 5, 1, 8, (1, 1)),226      # pylint: enable=bad-whitespace227  )228  activation: str = 'relu'229  batch_norm: str = 'default'230  bn_momentum: float = 0.99231  bn_epsilon: float = 1e-3232  # While the original implementation used a weight decay of 1e-5,233  # tf.nn.l2_loss divides it by 2, so we halve this to compensate in Keras234  weight_decay: float = 5e-6235  drop_connect_rate: float = 0.1236  depth_divisor: int = 8237  min_depth: Optional[int] = None238  # No Squeeze/Excite for MobilenetEdgeTPUV2239  use_se: bool = False240  input_channels: int = 3241  num_classes: int = 1001242  model_name: str = 'mobilenet_edgetpu_v2'243  rescale_input: bool = False244  data_format: str = 'channels_last'245  dtype: str = 'float32'246  # The number of filters in each group. HW arch dependent.247  group_base_size: int = 64248  backbone_only: bool = False249  features_as_dict: bool = False250def mobilenet_edgetpu_v2_base(251    width_coefficient: float = 1.0,252    depth_coefficient: float = 1.0,253    stem_base_filters: int = 64,254    stem_kernel_size: int = 5,255    top_base_filters: int = 1280,256    group_base_size: int = 64,257    dropout_rate: float = 0.2,258    drop_connect_rate: float = 0.1,259    filter_size_overrides: Optional[Dict[int, int]] = None,260    block_op_overrides: Optional[Dict[int, Dict[int, Dict[str, Any]]]] = None,261    block_group_overrides: Optional[Dict[int, Dict[str, Any]]] = None):262  """Creates MobilenetEdgeTPUV2 ModelConfig based on tuning parameters."""263  config = ModelConfig()264  param_overrides = {265      'width_coefficient': width_coefficient,266      'depth_coefficient': depth_coefficient,267      'stem_base_filters': stem_base_filters,268      'stem_kernel_size': stem_kernel_size,269      'top_base_filters': top_base_filters,270      'group_base_size': group_base_size,271      'dropout_rate': dropout_rate,272      'drop_connect_rate': drop_connect_rate273  }274  config = config.replace(**param_overrides)275  topology_config = TopologyConfig()276  if filter_size_overrides:277    for group_id in filter_size_overrides:278      topology_config.block_groups[group_id].filters = filter_size_overrides[279          group_id]280  if block_op_overrides:281    for group_id in block_op_overrides:282      for block_id in block_op_overrides[group_id]:283        replaced_block = topology_config.block_groups[group_id].blocks[284            block_id].replace(**block_op_overrides[group_id][block_id])285        topology_config.block_groups[group_id].blocks[block_id] = replaced_block286  if block_group_overrides:287    for group_id in block_group_overrides:288      replaced_group = topology_config.block_groups[group_id].replace(289          **block_group_overrides[group_id])290      topology_config.block_groups[group_id] = replaced_group291  blocks = ()292  input_filters = stem_base_filters293  for group in topology_config.block_groups:294    for block_search in group.blocks:295      if block_search.op_type != BlockType.skip:296        block = BlockConfig.from_search_config(297            input_filters=input_filters,298            output_filters=group.filters,299            block_search_config=block_search)300        blocks += (block,)301        # Set input filters for the next block302        input_filters = group.filters303  config = config.replace(blocks=blocks)304  return config305def autoseg_edgetpu_backbone_base(306    width_coefficient: float = 1.0,307    depth_coefficient: float = 1.0,308    stem_base_filters: int = 64,309    stem_kernel_size: int = 5,310    top_base_filters: int = 1280,311    group_base_size: int = 64,312    dropout_rate: float = 0.2,313    drop_connect_rate: float = 0.1,314    blocks_overrides: Optional[Tuple[BlockConfig, ...]] = None):315  """Creates a edgetpu ModelConfig based on search on segmentation."""316  config = ModelConfig()317  config.depth_divisor = 4318  param_overrides = {319      'width_coefficient': width_coefficient,320      'depth_coefficient': depth_coefficient,321      'stem_base_filters': stem_base_filters,322      'stem_kernel_size': stem_kernel_size,323      'top_base_filters': top_base_filters,324      'group_base_size': group_base_size,325      'dropout_rate': dropout_rate,326      'drop_connect_rate': drop_connect_rate,327  }328  if blocks_overrides:329    param_overrides['blocks'] = blocks_overrides330  config = config.replace(**param_overrides)331  return config332def autoseg_edgetpu_backbone_s() -> ModelConfig:333  """AutoML searched model with 2.5ms target simulated latency."""334  stem_base_filters = 32335  stem_kernel_size = 3336  top_base_filters = 1280337  blocks = (338      # (input_filters, output_filters, kernel_size, num_repeat,339      #  expand_ratio, strides, se_ratio, id_skip, fused_conv, conv_type)340      # pylint: disable=bad-whitespace341      BlockConfig.from_args(342          stem_base_filters,343          12,344          3,345          1,346          1, (1, 1),347          fused_expand=True,348          conv_type='full'),349      BlockConfig.from_args(350          12, 36, 3, 1, 6, (2, 2), fused_expand=True, conv_type='full'),351      BlockConfig.from_args(36, 18, 5, 1, 3, (1, 1)),352      BlockConfig.from_args(353          18, 60, 5, 1, 6, (2, 2), fused_expand=True, conv_type='full'),354      BlockConfig.from_args(60, 60, 3, 1, 3, (1, 1)),355      BlockConfig.from_args(60, 120, 5, 1, 6, (2, 2)),356      BlockConfig.from_args(120, 120, 3, 1, 3, (1, 1)),357      BlockConfig.from_args(120, 120, 5, 1, 6, (1, 1)),358      BlockConfig.from_args(120, 112, 3, 1, 6, (1, 1)),359      BlockConfig.from_args(112, 112, 5, 2, 6, (1, 1)),360      BlockConfig.from_args(112, 112, 5, 1, 1, (2, 2), id_skip=False),361      BlockConfig.from_args(362          112, 192, 1, 1, 6, (1, 1), fused_expand=True, id_skip=False),363      BlockConfig.from_args(192, 192, 5, 1, 1, (1, 1), id_skip=False),364      BlockConfig.from_args(365          192, 96, 1, 1, 6, (1, 1), fused_expand=True, id_skip=False),366      BlockConfig.from_args(96, 96, 5, 1, 3, (1, 1)),367      BlockConfig.from_args(96, 96, 5, 1, 1, (1, 1), id_skip=False),368      BlockConfig.from_args(369          96, 192, 1, 1, 6, (1, 1), fused_expand=True, id_skip=False),370      BlockConfig.from_args(192, 192, 5, 1, 1, (1, 1), id_skip=False),371      BlockConfig.from_args(372          192, 160, 1, 1, 3, (1, 1), fused_expand=True, id_skip=False),373      # pylint: enable=bad-whitespace374  )375  return autoseg_edgetpu_backbone_base(376      stem_base_filters=stem_base_filters,377      stem_kernel_size=stem_kernel_size,378      top_base_filters=top_base_filters,379      blocks_overrides=blocks,380      dropout_rate=0.2,381      drop_connect_rate=0.2)382def autoseg_edgetpu_backbone_xs() -> ModelConfig:383  """AutoML searched model with 2ms target simulated latency."""384  stem_base_filters = 32385  stem_kernel_size = 3386  top_base_filters = 1280387  blocks = (388      # (input_filters, output_filters, kernel_size, num_repeat,389      #  expand_ratio, strides, se_ratio, id_skip, fused_conv, conv_type)390      # pylint: disable=bad-whitespace391      BlockConfig.from_args(392          stem_base_filters,393          12,394          3,395          1,396          1, (1, 1),397          fused_expand=True,398          conv_type='full'),399      BlockConfig.from_args(400          12, 24, 3, 1, 6, (2, 2), fused_expand=True, conv_type='full'),401      BlockConfig.from_args(24, 24, 3, 1, 3, (1, 1)),402      BlockConfig.from_args(403          24, 60, 3, 1, 3, (2, 2), fused_expand=True, conv_type='full'),404      BlockConfig.from_args(60, 40, 3, 1, 6, (1, 1)),405      BlockConfig.from_args(40, 40, 5, 1, 3, (2, 2)),406      BlockConfig.from_args(40, 40, 3, 1, 6, (1, 1)),407      BlockConfig.from_args(408          40, 120, 3, 1, 6, (1, 1), fused_expand=True, conv_type='full'),409      BlockConfig.from_args(120, 168, 3, 1, 6, (1, 1)),410      BlockConfig.from_args(168, 84, 5, 1, 6, (1, 1)),411      BlockConfig.from_args(84, 84, 5, 1, 3, (1, 1)),412      BlockConfig.from_args(84, 84, 5, 1, 1, (2, 2), id_skip=False),413      BlockConfig.from_args(414          84, 288, 1, 1, 6, (1, 1), fused_expand=True, id_skip=False),415      BlockConfig.from_args(288, 288, 5, 1, 1, (1, 1), id_skip=False),416      BlockConfig.from_args(417          288, 96, 1, 1, 3, (1, 1), fused_expand=True, id_skip=False),418      BlockConfig.from_args(96, 96, 5, 1, 1, (1, 1), id_skip=False),419      BlockConfig.from_args(420          96, 96, 1, 1, 6, (1, 1), fused_expand=True, id_skip=False),421      BlockConfig.from_args(96, 96, 5, 1, 1, (1, 1), id_skip=False),422      BlockConfig.from_args(423          96, 96, 1, 1, 6, (1, 1), fused_expand=True, id_skip=False),424      BlockConfig.from_args(96, 480, 5, 1, 3, (1, 1)),425      # pylint: enable=bad-whitespace426  )427  return autoseg_edgetpu_backbone_base(428      stem_base_filters=stem_base_filters,429      stem_kernel_size=stem_kernel_size,430      top_base_filters=top_base_filters,431      blocks_overrides=blocks,432      dropout_rate=0.2,433      drop_connect_rate=0.2)434def autoseg_edgetpu_backbone_m() -> ModelConfig:435  """AutoML searched model with 3ms target simulated latency."""436  stem_base_filters = 32437  stem_kernel_size = 3438  top_base_filters = 1280439  blocks = (440      # (input_filters, output_filters, kernel_size, num_repeat,441      #  expand_ratio, strides, se_ratio, id_skip, fused_conv, conv_type)442      # pylint: disable=bad-whitespace443      BlockConfig.from_args(stem_base_filters, 16, 5, 1, 1, (1, 1)),444      BlockConfig.from_args(445          16, 36, 3, 1, 6, (2, 2), fused_expand=True, conv_type='full'),446      BlockConfig.from_args(36, 36, 3, 1, 3, (1, 1)),447      BlockConfig.from_args(448          36, 60, 3, 1, 6, (2, 2), fused_expand=True, conv_type='full'),449      BlockConfig.from_args(60, 60, 3, 1, 6, (1, 1)),450      BlockConfig.from_args(451          60, 120, 5, 1, 6, (2, 2), fused_expand=True, conv_type='full'),452      BlockConfig.from_args(120, 120, 5, 1, 6, (1, 1)),453      BlockConfig.from_args(454          120, 80, 3, 1, 6, (1, 1), fused_expand=True, conv_type='full'),455      BlockConfig.from_args(80, 168, 3, 1, 6, (1, 1)),456      BlockConfig.from_args(168, 168, 5, 1, 6, (1, 1)),457      BlockConfig.from_args(168, 168, 5, 1, 1, (1, 1), id_skip=False),458      BlockConfig.from_args(459          168, 168, 1, 1, 6, (1, 1), fused_expand=True, id_skip=False),460      BlockConfig.from_args(168, 168, 3, 1, 1, (2, 2), id_skip=False),461      BlockConfig.from_args(462          168, 192, 1, 1, 3, (1, 1), fused_expand=True, id_skip=False),463      BlockConfig.from_args(192, 192, 5, 1, 1, (1, 1), id_skip=False),464      BlockConfig.from_args(465          192, 288, 1, 1, 6, (1, 1), fused_expand=True, id_skip=False),466      BlockConfig.from_args(288, 288, 5, 1, 1, (1, 1), id_skip=False),467      BlockConfig.from_args(468          288, 96, 1, 1, 6, (1, 1), fused_expand=True, id_skip=False),469      BlockConfig.from_args(96, 96, 5, 1, 1, (1, 1), id_skip=False),470      BlockConfig.from_args(471          96, 192, 1, 1, 3, (1, 1), fused_expand=True, id_skip=False),472      BlockConfig.from_args(192, 192, 5, 1, 1, (1, 1), id_skip=False),473      BlockConfig.from_args(474          192, 320, 1, 1, 3, (1, 1), fused_expand=True, id_skip=False),475      # pylint: enable=bad-whitespace476  )477  return autoseg_edgetpu_backbone_base(478      stem_base_filters=stem_base_filters,479      stem_kernel_size=stem_kernel_size,480      top_base_filters=top_base_filters,481      blocks_overrides=blocks,482      dropout_rate=0.3,483      drop_connect_rate=0.3)484def mobilenet_edgetpu_v2_tiny() -> ModelConfig:485  """MobilenetEdgeTPUV2 tiny model config."""486  stem_base_filters = 32487  stem_kernel_size = 5488  top_base_filters = 1280489  filter_sizes = [16, 32, 48, 80, 112, 160, 192]490  filter_size_overrides = {491      k: v for (k, v) in zip(range(len(filter_sizes)), filter_sizes)492  }493  block_op_overrides = {494      2: {495          0: {'op_type': BlockType.from_args('ibn_fused_grouped')},496          2: {'op_type': BlockType.from_args('ibn_fused_grouped')},497      },498      3: {499          0: {'op_type': BlockType.from_args('ibn_fused_grouped')},500      }501  }502  return mobilenet_edgetpu_v2_base(503      stem_base_filters=stem_base_filters,504      stem_kernel_size=stem_kernel_size,505      top_base_filters=top_base_filters,506      filter_size_overrides=filter_size_overrides,507      block_op_overrides=block_op_overrides,508      dropout_rate=0.05,509      drop_connect_rate=0.05)510def mobilenet_edgetpu_v2_xs() -> ModelConfig:511  """MobilenetEdgeTPUV2 extra small model config."""512  stem_base_filters = 32513  stem_kernel_size = 5...
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.