rnnt_greedy_decoding.py: typos? auto-repressively -> auto-regressively (

NVIDIA#6989)

Signed-off-by: Vadim Kantorov <vadimkantorov@gmail.com>

nemo/collections/asr/parts/submodules/rnnt_greedy_decoding.py

@@ -241,7 +241,7 @@ def _joint_step(self, enc, pred, log_normalize: Optional[bool] = None):
 class GreedyRNNTInfer(_GreedyRNNTInfer):
  """A greedy transducer decoder.
 
- Sequence level greedy decoding, performed auto-repressively.
+ Sequence level greedy decoding, performed auto-regressively.
 
  Args:
  decoder_model: rnnt_utils.AbstractRNNTDecoder implementation.
@@ -326,7 +326,7 @@ def forward(
  partial_hypotheses: Optional[List[rnnt_utils.Hypothesis]] = None,
  ):
  """Returns a list of hypotheses given an input batch of the encoder hidden embedding.
- Output token is generated auto-repressively.
+ Output token is generated auto-regressively.
 
  Args:
  encoder_output: A tensor of size (batch, features, timesteps).
@@ -479,7 +479,7 @@ def _greedy_decode(
 class GreedyBatchedRNNTInfer(_GreedyRNNTInfer):
  """A batch level greedy transducer decoder.
 
- Batch level greedy decoding, performed auto-repressively.
+ Batch level greedy decoding, performed auto-regressively.
 
  Args:
  decoder_model: rnnt_utils.AbstractRNNTDecoder implementation.
@@ -571,7 +571,7 @@ def forward(
  partial_hypotheses: Optional[List[rnnt_utils.Hypothesis]] = None,
  ):
  """Returns a list of hypotheses given an input batch of the encoder hidden embedding.
- Output token is generated auto-repressively.
+ Output token is generated auto-regressively.
 
  Args:
  encoder_output: A tensor of size (batch, features, timesteps).
@@ -1034,7 +1034,7 @@ def __init__(self, encoder_model: str, decoder_joint_model: str, max_symbols_per
 
  def __call__(self, audio_signal: torch.Tensor, length: torch.Tensor):
  """Returns a list of hypotheses given an input batch of the encoder hidden embedding.
- Output token is generated auto-repressively.
+ Output token is generated auto-regressively.
 
  Args:
  encoder_output: A tensor of size (batch, features, timesteps).
@@ -1455,7 +1455,7 @@ def _get_initial_states(self, batchsize):
 class GreedyMultiblankRNNTInfer(GreedyRNNTInfer):
  """A greedy transducer decoder for multi-blank RNN-T.
 
- Sequence level greedy decoding, performed auto-repressively.
+ Sequence level greedy decoding, performed auto-regressively.
 
  Args:
  decoder_model: rnnt_utils.AbstractRNNTDecoder implementation.
@@ -1655,7 +1655,7 @@ def _greedy_decode(
 
 class GreedyBatchedMultiblankRNNTInfer(GreedyBatchedRNNTInfer):
  """A batch level greedy transducer decoder.
- Batch level greedy decoding, performed auto-repressively.
+ Batch level greedy decoding, performed auto-regressively.
  Args:
  decoder_model: rnnt_utils.AbstractRNNTDecoder implementation.
  joint_model: rnnt_utils.AbstractRNNTJoint implementation.
@@ -2207,7 +2207,7 @@ class GreedyBatchedRNNTInferConfig:
 class GreedyTDTInfer(_GreedyRNNTInfer):
  """A greedy TDT decoder.
 
- Sequence level greedy decoding, performed auto-repressively.
+ Sequence level greedy decoding, performed auto-regressively.
 
  Args:
  decoder_model: rnnt_utils.AbstractRNNTDecoder implementation.
@@ -2289,7 +2289,7 @@ def forward(
  partial_hypotheses: Optional[List[rnnt_utils.Hypothesis]] = None,
  ):
  """Returns a list of hypotheses given an input batch of the encoder hidden embedding.
- Output token is generated auto-repressively.
+ Output token is generated auto-regressively.
  Args:
  encoder_output: A tensor of size (batch, features, timesteps).
  encoded_lengths: list of int representing the length of each sequence
@@ -2459,7 +2459,7 @@ def _greedy_decode(
 
 class GreedyBatchedTDTInfer(_GreedyRNNTInfer):
  """A batch level greedy TDT decoder.
- Batch level greedy decoding, performed auto-repressively.
+ Batch level greedy decoding, performed auto-regressively.
  Args:
  decoder_model: rnnt_utils.AbstractRNNTDecoder implementation.
  joint_model: rnnt_utils.AbstractRNNTJoint implementation.
@@ -2547,7 +2547,7 @@ def forward(
  partial_hypotheses: Optional[List[rnnt_utils.Hypothesis]] = None,
  ):
  """Returns a list of hypotheses given an input batch of the encoder hidden embedding.
- Output token is generated auto-repressively.
+ Output token is generated auto-regressively.
  Args:
  encoder_output: A tensor of size (batch, features, timesteps).
  encoded_lengths: list of int representing the length of each sequence