|
|
@@ -31,7 +31,7 @@ llm: !new:cosyvoice.llm.llm.TransformerLM
|
|
|
num_blocks: 3
|
|
|
dropout_rate: 0.1
|
|
|
positional_dropout_rate: 0.1
|
|
|
- attention_dropout_rate: 0
|
|
|
+ attention_dropout_rate: 0.0
|
|
|
normalize_before: True
|
|
|
input_layer: 'linear'
|
|
|
pos_enc_layer_type: 'rel_pos_espnet'
|
|
|
@@ -49,7 +49,7 @@ llm: !new:cosyvoice.llm.llm.TransformerLM
|
|
|
num_blocks: 7
|
|
|
dropout_rate: 0.1
|
|
|
positional_dropout_rate: 0.1
|
|
|
- attention_dropout_rate: 0
|
|
|
+ attention_dropout_rate: 0.0
|
|
|
input_layer: 'linear_legacy'
|
|
|
pos_enc_layer_type: 'rel_pos_espnet'
|
|
|
selfattention_layer_type: 'rel_selfattn'
|
|
|
@@ -97,7 +97,7 @@ flow: !new:cosyvoice.flow.flow.MaskedDiffWithXvec
|
|
|
in_channels: 320
|
|
|
out_channels: 80
|
|
|
channels: [256, 256]
|
|
|
- dropout: 0
|
|
|
+ dropout: 0.0
|
|
|
attention_head_dim: 64
|
|
|
n_blocks: 4
|
|
|
num_mid_blocks: 8
|
tyanz