From ce3f59d9c7d0c9fda923aee3b38d101a00ecfb45 Mon Sep 17 00:00:00 2001 From: Daniel Povey Date: Thu, 22 Sep 2022 19:18:50 +0800 Subject: [PATCH] Use dropout in attention, on attn weights. --- egs/librispeech/ASR/pruned_transducer_stateless7/conformer.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/egs/librispeech/ASR/pruned_transducer_stateless7/conformer.py b/egs/librispeech/ASR/pruned_transducer_stateless7/conformer.py index a6ecef1e7..cf3129df2 100644 --- a/egs/librispeech/ASR/pruned_transducer_stateless7/conformer.py +++ b/egs/librispeech/ASR/pruned_transducer_stateless7/conformer.py @@ -168,7 +168,7 @@ class ConformerEncoderLayer(nn.Module): self.d_model = d_model self.self_attn = RelPositionMultiheadAttention( - d_model, nhead, dropout=0.0 + d_model, nhead, dropout=dropout, ) self.feed_forward = nn.Sequential(