def __init__(
self, vocab, beam_width, alpha, beta, lm_path, num_cpus, cutoff_prob=1.0, cutoff_top_n=40, input_tensor=True
):
try:
from ctc_decoders import Scorer
from ctc_decoders import ctc_beam_search_decoder_batch
except ModuleNotFoundError:
raise ModuleNotFoundError(
"BeamSearchDecoderWithLM requires the "
"installation of ctc_decoders "
"from nemo/scripts/install_decoders.py"
)
super().__init__()
# Override the default placement from neural factory and set placement/device to be CPU.
self._placement = DeviceType.CPU
self._device = get_cuda_device(self._placement)
if self._factory.world_size > 1:
raise ValueError("BeamSearchDecoderWithLM does not run in distributed mode")
self.scorer = Scorer(alpha, beta, model_path=lm_path, vocabulary=vocab)
self.beam_search_func = ctc_beam_search_decoder_batch
self.vocab = vocab
self.beam_width = beam_width
self.num_cpus = num_cpus
self.cutoff_prob = cutoff_prob
self.cutoff_top_n = cutoff_top_n
self.input_tensor = input_tensor
def __init__(self, name=None):
NeuralModule.__init__(self, name) # For NeuralModule API
# Set module type.
self._type = ModuleType.loss
self._device = get_cuda_device(self.placement)
def __init__(self, pretrained_model_name=None, name=None):
NeuralModule.__init__(self, name) # For NeuralModule API
nn.Module.__init__(self) # For PyTorch API
self._device = get_cuda_device(self.placement)
# Store pretrained model name (to be removed/changed)
self._pretrained_model_name = pretrained_model_name
def __init__(self, pretrained_model_name=None, name=None):
# Initialize nn.Module first - important for the inspect during the init_params collection.
nn.Module.__init__(self) # For PyTorch API
NeuralModule.__init__(self, name) # For NeuralModule API
# Set module type.
self._type = ModuleType.trainable
self._device = get_cuda_device(self.placement)
# Store pretrained model name (to be removed/changed)
self._pretrained_model_name = pretrained_model_name
def __init__(self):
NeuralModule.__init__(self) # For NeuralModule API
self._device = get_cuda_device(self.placement)
# if 'batch_size' not in kwargs:
# logging.warning("No batch_size specified in the data layer. "
# "Setting batch_size to 1.")
# kwargs['batch_size'] = 1
# Set default values of variables used by trained/passed to DataLoader.
# NOTE: That also means that those are parameters of DataLoader/trainer, not DataLayer.
# Thus those fields will be removed from DataLayer and moved to trainer configuration
# (when the time for that will come;))
self._batch_size = 1
self._num_workers = os.cpu_count() # Use all CPUs by default.
self._shuffle = False # Don't shuffle by default.
def __init__(self):
NeuralModule.__init__(self) # For NeuralModule API
self._device = get_cuda_device(self.placement)
