Increase layerdrop_prob of ConvNeXt, and make it warm up faster.

This commit is contained in:
Daniel Povey 2022-12-17 16:44:57 +08:00
parent 286b2021c2
commit 111a0aa3c7

View File

@ -1680,7 +1680,7 @@ class ConvNeXt(nn.Module):
pad = (kernel_size - 1) // 2 pad = (kernel_size - 1) // 2
hidden_channels = channels * hidden_ratio hidden_channels = channels * hidden_ratio
if layerdrop_prob is None: if layerdrop_prob is None:
layerdrop_prob = ScheduledFloat((0.0, 0.1), (16000.0, 0.01)) layerdrop_prob = ScheduledFloat((0.0, 0.2), (4000.0, 0.025))
self.layerdrop_prob = layerdrop_prob self.layerdrop_prob = layerdrop_prob
self.depthwise_conv = nn.Conv2d( self.depthwise_conv = nn.Conv2d(