/
custom.py
68 lines (58 loc) · 2.74 KB
/
custom.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
import chainer
import torch
from espnet.nets.asr_interface import ASRInterface
from espnet.nets.pytorch_backend.transformer.encoder import Encoder
from espnet.nets.pytorch_backend.transformer.decoder import Decoder
from espnet.nets.pytorch_backend.transformer.mask import subsequent_mask
from espnet.nets.pytorch_backend.transformer.label_smoothing_loss import LabelSmoothingLoss
from espnet.nets.pytorch_backend.nets_utils import make_pad_mask
from espnet.nets.pytorch_backend.nets_utils import th_accuracy
class Reporter(chainer.Chain):
def report(self, **kwargs):
chainer.reporter.report(kwargs, self)
class ASRTransformer(ASRInterface, torch.nn.Module):
@staticmethod
def add_arguments(parser):
parser.add_argument("--label-smoothing", default=0.0, type=float)
return parser
def __init__(self, idim, odim, args=None):
torch.nn.Module.__init__(self)
self.encoder = Encoder(idim, input_layer="linear")
self.decoder = Decoder(odim)
self.criterion = LabelSmoothingLoss(odim, -1, args.label_smoothing, True)
self.sos = odim - 1
self.eos = odim - 1
self.ignore_id=-1
self.subsample = [0]
self.reporter = Reporter()
# for training
def forward(self, xs_pad, ilens, ys_pad):
"""Compute scalar loss for backprop"""
src_mask = (~make_pad_mask(ilens.tolist())).to(xs_pad.device).unsqueeze(-2)
hs_pad, hs_mask = self.encoder(xs_pad, src_mask)
ys_in_pad, ys_out_pad = self.add_sos_eos(ys_pad)
ys_mask = self.target_mask(ys_in_pad)
pred_pad, pred_mask = self.decoder(ys_in_pad, ys_mask, hs_pad, hs_mask)
loss = self.criterion(pred_pad, ys_out_pad)
self.acc = th_accuracy(pred_pad.view(-1, pred_pad.size(-1)), ys_out_pad, ignore_label=self.ignore_id)
self.reporter.report(loss=loss, acc=self.acc)
return loss
def add_sos_eos(self, ys_pad):
from espnet.nets.pytorch_backend.nets_utils import pad_list
eos = ys_pad.new([self.eos])
sos = ys_pad.new([self.sos])
ys = [y[y != self.ignore_id] for y in ys_pad] # parse padded ys
ys_in = [torch.cat([sos, y], dim=0) for y in ys]
ys_out = [torch.cat([y, eos], dim=0) for y in ys]
return pad_list(ys_in, self.eos), pad_list(ys_out, self.ignore_id)
def target_mask(self, ys_in_pad):
ys_mask = ys_in_pad != self.ignore_id
m = subsequent_mask(ys_mask.size(-1), device=ys_mask.device).unsqueeze(0)
return ys_mask.unsqueeze(-2) & m
# for decoding
def encode(self, feat):
"""Encode speech feature."""
return self.encoder(feat.unsqueeze(0), None)[0][0]
def scorers(self):
"""Scorer used in beam search"""
return {"decoder": self.decoder}