mirrors/icefall
1
0
Fork 0
mirror of https://github.com/k2-fsa/icefall.git synced 2025-12-11 06:55:27 +00:00
Code Issues Packages Projects Releases Wiki Activity

Modify feature_mask_dropout_prob

Browse Source
This commit is contained in:
Daniel Povey 2023-03-31 13:25:39 +08:00
parent e64ec396bd
commit d41b73000e
1 changed files with 5 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
egs/librispeech/ASR/pruned_transducer_stateless7/zipformer.py
View File

@ -297,21 +297,20 @@ class Zipformer2(EncoderInterface):
num_frames_max = (num_frames0 + max_downsampling_factor - 1)
# we divide the dropped-out feature dimensions into two equal groups;
# the first group is dropped out with probability 0.05, the second
# with probability approximately (0.2 + 0.05).
feature_mask_dropout_prob1 = 0.05
feature_mask_dropout_prob2 = 0.2
# the first group is dropped out with probability 0.1, the second
# with probability approximately twice that.
feature_mask_dropout_prob = 0.1
# frame_mask_max1 shape: (num_frames_max, batch_size, 1)
frame_mask_max1 = (torch.rand(num_frames_max, batch_size, 1,
device=x.device) >
feature_mask_dropout_prob1).to(x.dtype)
feature_mask_dropout_prob).to(x.dtype)
# frame_mask_max2 has additional frames masked, about twice the number.
frame_mask_max2 = torch.logical_and(frame_mask_max1,
(torch.rand(num_frames_max, batch_size, 1,
device=x.device) >
feature_mask_dropout_prob2).to(x.dtype))
feature_mask_dropout_prob).to(x.dtype))
# dim: (num_frames_max, batch_size, 3)