cemsubakan
commited on
Commit
•
ecc66bd
1
Parent(s):
2e27224
Update hyperparams.yaml
Browse files- hyperparams.yaml +11 -11
hyperparams.yaml
CHANGED
@@ -3,7 +3,7 @@
|
|
3 |
# Dataset : WHAM!
|
4 |
# ################################
|
5 |
|
6 |
-
num_spks: 1
|
7 |
sample_rate: 8000
|
8 |
|
9 |
# Encoder parameters
|
@@ -13,11 +13,11 @@ kernel_size: 16
|
|
13 |
kernel_stride: 8
|
14 |
|
15 |
# Specifying the network
|
16 |
-
Encoder:
|
17 |
kernel_size: 16
|
18 |
out_channels: 256
|
19 |
|
20 |
-
SBtfintra:
|
21 |
num_layers: 8
|
22 |
d_model: 256
|
23 |
nhead: 8
|
@@ -26,7 +26,7 @@ SBtfintra: &id001 !new:speechbrain.lobes.models.dual_path.SBTransformerBlock
|
|
26 |
use_positional_encoding: true
|
27 |
norm_before: true
|
28 |
|
29 |
-
SBtfinter:
|
30 |
num_layers: 8
|
31 |
d_model: 256
|
32 |
nhead: 8
|
@@ -35,19 +35,19 @@ SBtfinter: &id002 !new:speechbrain.lobes.models.dual_path.SBTransformerBlock
|
|
35 |
use_positional_encoding: true
|
36 |
norm_before: true
|
37 |
|
38 |
-
MaskNet:
|
39 |
num_spks: 1
|
40 |
in_channels: 256
|
41 |
out_channels: 256
|
42 |
num_layers: 2
|
43 |
K: 250
|
44 |
-
intra_model:
|
45 |
-
inter_model:
|
46 |
norm: ln
|
47 |
linear_layer_after_inter_intra: false
|
48 |
skip_around_intra: true
|
49 |
|
50 |
-
Decoder:
|
51 |
in_channels: 256
|
52 |
out_channels: 1
|
53 |
kernel_size: 16
|
@@ -55,9 +55,9 @@ Decoder: &id004 !new:speechbrain.lobes.models.dual_path.Decoder
|
|
55 |
bias: false
|
56 |
|
57 |
modules:
|
58 |
-
encoder:
|
59 |
-
decoder:
|
60 |
-
masknet:
|
61 |
|
62 |
pretrainer: !new:speechbrain.utils.parameter_transfer.Pretrainer
|
63 |
loadables:
|
|
|
3 |
# Dataset : WHAM!
|
4 |
# ################################
|
5 |
|
6 |
+
num_spks: 1
|
7 |
sample_rate: 8000
|
8 |
|
9 |
# Encoder parameters
|
|
|
13 |
kernel_stride: 8
|
14 |
|
15 |
# Specifying the network
|
16 |
+
Encoder: !new:speechbrain.lobes.models.dual_path.Encoder
|
17 |
kernel_size: 16
|
18 |
out_channels: 256
|
19 |
|
20 |
+
SBtfintra: !new:speechbrain.lobes.models.dual_path.SBTransformerBlock
|
21 |
num_layers: 8
|
22 |
d_model: 256
|
23 |
nhead: 8
|
|
|
26 |
use_positional_encoding: true
|
27 |
norm_before: true
|
28 |
|
29 |
+
SBtfinter: !new:speechbrain.lobes.models.dual_path.SBTransformerBlock
|
30 |
num_layers: 8
|
31 |
d_model: 256
|
32 |
nhead: 8
|
|
|
35 |
use_positional_encoding: true
|
36 |
norm_before: true
|
37 |
|
38 |
+
MaskNet: !new:speechbrain.lobes.models.dual_path.Dual_Path_Model
|
39 |
num_spks: 1
|
40 |
in_channels: 256
|
41 |
out_channels: 256
|
42 |
num_layers: 2
|
43 |
K: 250
|
44 |
+
intra_model: !ref <SBtfintra>
|
45 |
+
inter_model: !ref <SBtfinter>
|
46 |
norm: ln
|
47 |
linear_layer_after_inter_intra: false
|
48 |
skip_around_intra: true
|
49 |
|
50 |
+
Decoder: !new:speechbrain.lobes.models.dual_path.Decoder
|
51 |
in_channels: 256
|
52 |
out_channels: 1
|
53 |
kernel_size: 16
|
|
|
55 |
bias: false
|
56 |
|
57 |
modules:
|
58 |
+
encoder: !ref <Encoder>
|
59 |
+
decoder: !ref <Decoder>
|
60 |
+
masknet: !ref <MaskNet>
|
61 |
|
62 |
pretrainer: !new:speechbrain.utils.parameter_transfer.Pretrainer
|
63 |
loadables:
|