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

Increase initial whitening target for self_attn from 2.0 to 3.0.

Browse Source
This commit is contained in:
Daniel Povey 2022-11-24 15:18:28 +08:00
parent 45069175d9
commit 0ac26f4234
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
egs/librispeech/ASR/pruned_transducer_stateless7/zipformer.py
View File

@ -1055,7 +1055,7 @@ class RelPositionMultiheadAttentionWeights(nn.Module):
initial_scale=query_head_dim**-0.25) initial_scale=query_head_dim**-0.25)
self.whiten_keys = Whiten(num_groups=num_heads, self.whiten_keys = Whiten(num_groups=num_heads,
whitening_limit=_whitening_schedule(2.0), whitening_limit=_whitening_schedule(3.0),
prob=(0.025, 0.25), prob=(0.025, 0.25),
grad_scale=0.025) grad_scale=0.025)