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

Introduce balancer schedules for the DoubleSwish() in feedforward and conv modules

Browse Source
This commit is contained in:
Daniel Povey 2022-11-26 20:20:20 +08:00
parent 320c58401f
commit 4874ded2e9
1 changed files with 8 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
egs/librispeech/ASR/pruned_transducer_stateless7/zipformer.py
View File

@ -1391,7 +1391,9 @@ class FeedforwardModule(nn.Module):
aux_grad_scale=_aux_grad_scale(), prob=_aux_grad_prob())
self.hidden_balancer = ActivationBalancer(feedforward_dim,
channel_dim=-1, max_abs=10.0,
channel_dim=-1,
min_positive=ScheduledFloat((0.0, 0.1), (12000.0, 0.05)),
max_abs=ScheduledFloat((0.0, 4.0), (12000.0, 10.0), default=10),
min_prob=0.25)
self.activation = DoubleSwish()
self.dropout = nn.Dropout(dropout)
@ -1435,7 +1437,7 @@ class NonlinAttentionModule(nn.Module):
channels // 2, channel_dim=-1,
min_positive=0.05, max_positive=1.0,
min_abs=0.2, max_abs=ScheduledFloat((0.0, 2.0),
(4000.0, 10.0),
(12000.0, 10.0),
default=1.0),
)
self.sigmoid = nn.Sigmoid()
@ -1529,7 +1531,7 @@ class ConvolutionModule(nn.Module):
self.deriv_balancer1 = ActivationBalancer(
2 * channels, channel_dim=-1,
max_abs=ScheduledFloat((0.0, 2.0),
(4000.0, 10.0),
(12000.0, 10.0),
default=1.0),
min_positive=0.05, max_positive=1.0
)
@ -1549,8 +1551,9 @@ class ConvolutionModule(nn.Module):
self.deriv_balancer2 = ActivationBalancer(
channels, channel_dim=1,
min_positive=0.05, max_positive=1.0,
max_abs=20.0,
min_positive=ScheduledFloat((0.0, 0.1), (12000.0, 0.05)),
max_abs=ScheduledFloat((0.0, 4.0), (12000.0, 20.0), default=10),
max_positive=1.0,
)
self.activation = DoubleSwish()