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
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
|
#pragma once
#include "marian.h"
#include "models/s2s.h"
namespace marian {
class Nematus : public EncoderDecoder {
public:
Nematus(Ptr<ExpressionGraph> graph, Ptr<Options> options) : EncoderDecoder(graph, options), nameMap_(createNameMap()) {
ABORT_IF(options_->get<std::string>("enc-type") != "bidirectional",
"--type nematus does not support other encoder "
"type than bidirectional, use --type s2s");
ABORT_IF(options_->get<std::string>("enc-cell") != "gru-nematus",
"--type nematus does not support other rnn cells "
"than gru-nematus, use --type s2s");
ABORT_IF(options_->get<std::string>("dec-cell") != "gru-nematus",
"--type nematus does not support other rnn cells "
"than gru-nematus, use --type s2s");
ABORT_IF(options_->get<int>("dec-cell-high-depth") > 1,
"--type nematus does not currently support "
"--dec-cell-high-depth > 1, use --type s2s");
}
void load(Ptr<ExpressionGraph> graph,
const std::vector<io::Item>& items,
bool /*markReloaded*/ = true) override {
auto ioItems = items;
// map names and remove a dummy matrix 'decoder_c_tt' from items to avoid creating isolated node
for(auto it = ioItems.begin(); it != ioItems.end();) {
// for backwards compatibility, turn one-dimensional vector into two dimensional matrix with first dimension being 1 and second dimension of the original size
// @TODO: consider dropping support for Nematus models
if(it->shape.size() == 1) {
int dim = it->shape[-1];
it->shape.resize(2);
it->shape.set(0, 1);
it->shape.set(1, dim);
}
if(it->name == "decoder_c_tt") {
it = ioItems.erase(it);
} else if(it->name == "uidx") {
it = ioItems.erase(it);
} else if(it->name == "history_errs") {
it = ioItems.erase(it);
} else {
auto pair = nameMap_.find(it->name);
if(pair != nameMap_.end())
it->name = pair->second;
it++;
}
}
// load items into the graph
graph->load(ioItems);
}
void load(Ptr<ExpressionGraph> graph,
const std::string& name,
bool /*markReloaded*/ = true) override {
LOG(info, "Loading model from {}", name);
auto ioItems = io::loadItems(name);
load(graph, ioItems);
}
void save(Ptr<ExpressionGraph> graph,
const std::string& name,
bool saveTranslatorConfig = false) override {
LOG(info, "Saving model to {}", name);
// prepare reversed map
if(nameMapRev_.empty())
for(const auto& kv : nameMap_)
nameMapRev_.insert({kv.second, kv.first});
// get parameters from the graph to items
std::vector<io::Item> ioItems;
graph->save(ioItems);
// replace names to be compatible with Nematus
for(auto& item : ioItems) {
auto newItemName = nameMapRev_.find(item.name);
if(newItemName != nameMapRev_.end())
item.name = newItemName->second;
}
// add a dummy matrix 'decoder_c_tt' required for Amun and Nematus
ioItems.emplace_back();
ioItems.back().name = "decoder_c_tt";
ioItems.back().shape = Shape({1, 0});
ioItems.back().bytes.emplace_back((char)0);
io::addMetaToItems(getModelParametersAsString(), "special:model.yml", ioItems);
io::saveItems(name, ioItems);
if(saveTranslatorConfig) {
createAmunConfig(name);
createDecoderConfig(name);
}
}
private:
std::map<std::string, std::string> nameMap_;
std::map<std::string, std::string> nameMapRev_;
std::map<std::string, std::string> createNameMap() {
std::map<std::string, std::string> nameMap
= {{"decoder_U", "decoder_cell1_U"},
{"decoder_Ux", "decoder_cell1_Ux"},
{"decoder_W", "decoder_cell1_W"},
{"decoder_Wx", "decoder_cell1_Wx"},
{"decoder_b", "decoder_cell1_b"},
{"decoder_bx", "decoder_cell1_bx"},
{"decoder_U_nl", "decoder_cell2_U"},
{"decoder_Ux_nl", "decoder_cell2_Ux"},
{"decoder_Wc", "decoder_cell2_W"},
{"decoder_Wcx", "decoder_cell2_Wx"},
{"decoder_b_nl", "decoder_cell2_b"},
{"decoder_bx_nl", "decoder_cell2_bx"},
{"ff_logit_prev_W", "decoder_ff_logit_l1_W0"},
{"ff_logit_lstm_W", "decoder_ff_logit_l1_W1"},
{"ff_logit_ctx_W", "decoder_ff_logit_l1_W2"},
{"ff_logit_prev_b", "decoder_ff_logit_l1_b0"},
{"ff_logit_lstm_b", "decoder_ff_logit_l1_b1"},
{"ff_logit_ctx_b", "decoder_ff_logit_l1_b2"},
{"ff_logit_W", "decoder_ff_logit_l2_W"},
{"ff_logit_b", "decoder_ff_logit_l2_b"},
{"ff_state_W", "decoder_ff_state_W"},
{"ff_state_b", "decoder_ff_state_b"},
{"Wemb_dec", "decoder_Wemb"},
{"Wemb", "encoder_Wemb"},
{"encoder_U", "encoder_bi_U"},
{"encoder_Ux", "encoder_bi_Ux"},
{"encoder_W", "encoder_bi_W"},
{"encoder_Wx", "encoder_bi_Wx"},
{"encoder_b", "encoder_bi_b"},
{"encoder_bx", "encoder_bi_bx"},
{"encoder_r_U", "encoder_bi_r_U"},
{"encoder_r_Ux", "encoder_bi_r_Ux"},
{"encoder_r_W", "encoder_bi_r_W"},
{"encoder_r_Wx", "encoder_bi_r_Wx"},
{"encoder_r_b", "encoder_bi_r_b"},
{"encoder_r_bx", "encoder_bi_r_bx"},
{"ff_state_ln_s", "decoder_ff_state_ln_s"},
{"ff_state_ln_b", "decoder_ff_state_ln_b"},
{"ff_logit_prev_ln_s", "decoder_ff_logit_l1_ln_s0"},
{"ff_logit_lstm_ln_s", "decoder_ff_logit_l1_ln_s1"},
{"ff_logit_ctx_ln_s", "decoder_ff_logit_l1_ln_s2"},
{"ff_logit_prev_ln_b", "decoder_ff_logit_l1_ln_b0"},
{"ff_logit_lstm_ln_b", "decoder_ff_logit_l1_ln_b1"},
{"ff_logit_ctx_ln_b", "decoder_ff_logit_l1_ln_b2"}};
// add mapping for deep encoder cells
std::vector<std::string> suffixes = {"_U", "_Ux", "_b", "_bx"};
for(int i = 1; i < options_->get<int>("enc-cell-depth"); ++i) {
std::string num1 = std::to_string(i);
std::string num2 = std::to_string(i + 1);
for(auto suf : suffixes) {
nameMap.insert({"encoder" + suf + "_drt_" + num1, "encoder_bi_cell" + num2 + suf});
nameMap.insert({"encoder_r" + suf + "_drt_" + num1, "encoder_bi_r_cell" + num2 + suf});
}
}
// add mapping for deep decoder cells
for(int i = 3; i <= options_->get<int>("dec-cell-base-depth"); ++i) {
std::string num1 = std::to_string(i - 2);
std::string num2 = std::to_string(i);
for(auto suf : suffixes)
nameMap.insert({"decoder" + suf + "_nl_drt_" + num1, "decoder_cell" + num2 + suf});
}
// add mapping for normalization layers
std::map<std::string, std::string> nameMapCopy(nameMap);
for(auto& kv : nameMapCopy) {
std::string prefix = kv.first.substr(0, 7);
if(prefix == "encoder" || prefix == "decoder") {
nameMap.insert({kv.first + "_lns", kv.second + "_lns"});
nameMap.insert({kv.first + "_lnb", kv.second + "_lnb"});
}
}
return nameMap;
}
void createAmunConfig(const std::string& name) {
Config::YamlNode amun;
// Amun has only CPU decoder for deep Nematus models
amun["cpu-threads"] = 16;
amun["gpu-threads"] = 0;
amun["maxi-batch"] = 1;
amun["mini-batch"] = 1;
auto vocabs = options_->get<std::vector<std::string>>("vocabs");
amun["source-vocab"] = vocabs[0];
amun["target-vocab"] = vocabs[1];
amun["devices"] = options_->get<std::vector<size_t>>("devices");
amun["normalize"] = true;
amun["beam-size"] = 5;
amun["relative-paths"] = false;
amun["scorers"]["F0"]["path"] = name;
amun["scorers"]["F0"]["type"] = "nematus2";
amun["weights"]["F0"] = 1.0f;
io::OutputFileStream out(name + ".amun.yml");
out << amun;
}
};
} // namespace marian
|
