ct2rs 0.9.19

Rust bindings for OpenNMT/CTranslate2
Documentation 
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

#include "ctranslate2/models/transformer.h"

#include "ctranslate2/layers/transformer.h"

namespace ctranslate2 {
  namespace models {

    static bool replace(std::string& str, const std::string& from, const std::string& to) {
      size_t start_pos = str.find(from);
      if (start_pos == std::string::npos)
        return false;
      str.replace(start_pos, from.length(), to);
      return true;
    }

    static std::string map_v1_variable_name(std::string name) {
      // V1 variable names were simply the names defined by OpenNMT-tf.
      replace(name, "transformer/", "");
      replace(name, ":0", "");
      replace(name, "w_embs", "embeddings/weight");
      replace(name, "kernel", "weight");
      replace(name, "LayerNorm", "layer_norm");
      replace(name, "dense", "projection");
      replace(name, "conv1d_", "linear_");
      replace(name, "conv1d", "linear_0");
      if (name.find("encoder") != std::string::npos) {
        replace(name, "multi_head", "self_attention");
      } else {
        replace(name, "masked_multi_head", "self_attention");
        replace(name, "multi_head", "attention");
      }
      return name;
    }


    TransformerModel::TransformerModel(size_t num_heads)
      : _num_heads(num_heads) {
    }

    size_t TransformerModel::current_spec_revision() const {
      return 7;
    }

    bool TransformerModel::is_linear_weight(const std::string& variable_name) const {
      // Linear weights are all variables that are quantizable and not under the "embeddings" scope.
      return is_quantizable(variable_name) && variable_name.find("embeddings") == std::string::npos;
    }

    bool TransformerModel::is_packable(const std::string& variable_name) const {
      // Disallow packing for the last linear layer if it can be dynamically masked.
      return (is_linear_weight(variable_name)
              && (!get_vocabulary_map() || variable_name.find("projection") == std::string::npos));
    }

    void TransformerModel::register_variable(std::string name, StorageView variable) {
      if (spec_revision() == 1)
        name = map_v1_variable_name(std::move(name));
      SequenceToSequenceModel::register_variable(std::move(name), std::move(variable));
    }

    void TransformerModel::initialize(ModelReader& model_reader) {
      SequenceToSequenceModel::initialize(model_reader);

      if (spec_revision() < 3) {
        register_variable("num_heads", StorageView(int8_t(_num_heads)));
      }

      if (spec_revision() < 5) {
        register_variable_alias("encoder/num_heads", "num_heads");
        register_variable_alias("encoder/pre_norm", "pre_norm");
        register_variable_alias("encoder/activation", "activation");
        register_variable_alias("encoder/embeddings_merge", "embeddings_merge");

        register_variable_alias("decoder/num_heads", "num_heads");
        register_variable_alias("decoder/pre_norm", "pre_norm");
        register_variable_alias("decoder/activation", "activation");
        register_variable_alias("decoder/alignment_layer", "alignment_layer");
        register_variable_alias("decoder/alignment_heads", "alignment_heads");
      }
    }

    std::unique_ptr<SequenceToSequenceReplica> TransformerModel::as_sequence_to_sequence() const {
      const auto scoped_device_setter = get_scoped_device_setter();

      auto encoder = std::make_unique<layers::TransformerEncoder>(*this, "encoder");
      auto decoder = std::make_unique<layers::TransformerDecoder>(*this, "decoder");

      const auto model = std::static_pointer_cast<const TransformerModel>(shared_from_this());
      return std::make_unique<EncoderDecoderReplica>(model, std::move(encoder), std::move(decoder));
    }

    std::unique_ptr<Model> TransformerModel::clone() const {
      return std::make_unique<TransformerModel>(*this);
    }


    size_t TransformerDecoderModel::current_spec_revision() const {
      return 8;
    }

    void TransformerDecoderModel::initialize(ModelReader& model_reader) {
      LanguageModel::initialize(model_reader);

      if (spec_revision() < 2) {
        register_variable_alias("decoder/num_heads", "num_heads");
        register_variable_alias("decoder/pre_norm", "pre_norm");
        register_variable_alias("decoder/activation", "activation");
      }
    }

    std::unique_ptr<SequenceGeneratorReplica>
    TransformerDecoderModel::as_sequence_generator() const {
      const auto scoped_device_setter = get_scoped_device_setter();

      auto decoder = std::make_unique<layers::TransformerDecoder>(*this, "decoder");

      const auto model = std::static_pointer_cast<const TransformerDecoderModel>(shared_from_this());
      return std::make_unique<DecoderReplica>(model, std::move(decoder));
    }

    bool TransformerDecoderModel::is_linear_weight(const std::string& variable_name) const {
      return is_quantizable(variable_name) && variable_name.find("embeddings") == std::string::npos;
    }

    std::unique_ptr<Model> TransformerDecoderModel::clone() const {
      return std::make_unique<TransformerDecoderModel>(*this);
    }


    size_t TransformerEncoderModel::current_spec_revision() const {
      return 1;
    }

    std::unique_ptr<SequenceEncoderReplica>
    TransformerEncoderModel::as_sequence_encoder() const {
      const auto scoped_device_setter = get_scoped_device_setter();

      auto encoder = std::make_unique<layers::TransformerEncoder>(*this, "encoder");

      const auto model = std::static_pointer_cast<const TransformerEncoderModel>(shared_from_this());
      return std::make_unique<EncoderReplica>(model, std::move(encoder));
    }

    bool TransformerEncoderModel::is_linear_weight(const std::string& variable_name) const {
      return is_quantizable(variable_name) && variable_name.find("embeddings") == std::string::npos;
    }

    std::unique_ptr<Model> TransformerEncoderModel::clone() const {
      return std::make_unique<TransformerEncoderModel>(*this);
    }

  }
}