minor fix for conformer subsampling docstring. (NVIDIA#7195)

Signed-off-by: Xuesong Yang <[email protected]> Signed-off-by: dorotat <[email protected]>
dorotat-nv · Aug 24, 2023 · 8aa94b9 · 8aa94b9
1 parent a5dc86c
commit 8aa94b9
Show file tree

Hide file tree

Showing 2 changed files with 2 additions and 2 deletions.
diff --git a/nemo/collections/asr/modules/conformer_encoder.py b/nemo/collections/asr/modules/conformer_encoder.py
@@ -65,7 +65,7 @@ class ConformerEncoder(NeuralModule, StreamingEncoder, Exportable, AccessMixin):
         d_model (int): the hidden size of the model
         feat_out (int): the size of the output features
             Defaults to -1 (means feat_out is d_model)
-        subsampling (str): the method of subsampling, choices=['vggnet', 'striding']
+        subsampling (str): the method of subsampling, choices=['vggnet', 'striding', 'dw-striding', 'stacking', 'stacking_norm']
             Defaults to striding.
         subsampling_factor (int): the subsampling factor which should be power of 2
             Defaults to 4.

diff --git a/nemo/collections/asr/parts/submodules/subsampling.py b/nemo/collections/asr/parts/submodules/subsampling.py
@@ -64,7 +64,7 @@ class ConvSubsampling(torch.nn.Module):
     VGGNet Subsampling: Transformer-transducer: end-to-end speech recognition with self-attention (https://arxiv.org/pdf/1910.12977.pdf)
     Striding Subsampling: "Speech-Transformer: A No-Recurrence Sequence-to-Sequence Model for Speech Recognition" by Linhao Dong et al. (https://ieeexplore.ieee.org/document/8462506)
     Args:
-        subsampling (str): The subsampling technique from {"vggnet", "striding"}
+        subsampling (str): The subsampling technique from {"vggnet", "striding", "dw-striding"}
         subsampling_factor (int): The subsampling factor which should be a power of 2
         subsampling_conv_chunking_factor (int): Input chunking factor which can be -1 (no chunking) 
         1 (auto) or a power of 2. Default is 1