am-nnet.cc
2.57 KB
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
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
// nnet2/am-nnet.cc
// Copyright 2012 Johns Hopkins University (author: Daniel Povey)
// See ../../COPYING for clarification regarding multiple authors
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// THIS CODE IS PROVIDED *AS IS* BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, EITHER EXPRESS OR IMPLIED, INCLUDING WITHOUT LIMITATION ANY IMPLIED
// WARRANTIES OR CONDITIONS OF TITLE, FITNESS FOR A PARTICULAR PURPOSE,
// MERCHANTABLITY OR NON-INFRINGEMENT.
// See the Apache 2 License for the specific language governing permissions and
// limitations under the License.
#include "nnet2/am-nnet.h"
namespace kaldi {
namespace nnet2 {
void AmNnet::Init(std::istream &config_is) {
nnet_.Init(config_is);
}
void AmNnet::Write(std::ostream &os, bool binary) const {
// We don't write any header or footer like <AmNnet> and </AmNnet> -- we just
// write the neural net and then the priors. Who knows, there might be some
// situation where we want to just read the neural net.
nnet_.Write(os, binary);
priors_.Write(os, binary);
}
void AmNnet::Read(std::istream &is, bool binary) {
nnet_.Read(is, binary);
priors_.Read(is, binary);
}
void AmNnet::SetPriors(const VectorBase<BaseFloat> &priors) {
priors_ = priors;
if (priors_.Dim() > NumPdfs())
KALDI_ERR << "Dimension of priors cannot exceed number of pdfs.";
if (priors_.Dim() > 0 && priors_.Dim() < NumPdfs()) {
KALDI_WARN << "Dimension of priors is " << priors_.Dim() << " < "
<< NumPdfs() << ": extending with zeros, in case you had "
<< "unseen pdf's, but this possibly indicates a serious problem.";
priors_.Resize(NumPdfs(), kCopyData);
}
}
std::string AmNnet::Info() const {
std::ostringstream ostr;
ostr << "prior dimension: " << priors_.Dim();
if (priors_.Dim() != 0) {
ostr << ", prior sum: " << priors_.Sum() << ", prior min: " << priors_.Min()
<< "\n";
}
return nnet_.Info() + ostr.str();
}
void AmNnet::Init(const Nnet &nnet) {
nnet_ = nnet;
if (priors_.Dim() != 0 && priors_.Dim() != nnet.OutputDim()) {
KALDI_WARN << "Initializing neural net: prior dimension mismatch, "
<< "discarding old priors.";
priors_.Resize(0);
}
}
void AmNnet::ResizeOutputLayer(int32 new_num_pdfs) {
nnet_.ResizeOutputLayer(new_num_pdfs);
priors_.Resize(new_num_pdfs);
priors_.Set(1.0 / new_num_pdfs);
}
} // namespace nnet2
} // namespace kaldi